Early detection of RVF is a prerequisite to effective control of the disease. Sentinel herd monitoring has been used in different parts of Africa to monitor viral circulation in susceptible populations. It can be enhanced by the additional monitoring of climatic parameters (see: Towards early warning for RVF prevention: satellite imagery, p. 35).
Activities should be directed towards active disease surveillance in order to build up baseline information on inter-epidemic virus transmission patterns, areas at risk and early warning of any increased virus activity or buildup in vector mosquito populations. This surveillance should be carried out by regular field visits and contact with livestock farmers and communities and should include periodic purposefully designed and geographically representative serological surveys and participatory epidemiological techniques. The detection of RVF virus activity by serology is usually too late to be of any relevance for control.
Sentinel herds are an important means of obtaining baseline epidemiological information on RVF. These are small ruminant herds located in geographically representative areas. Locations where mosquito breeding activity is likely to be greatest, e.g. near rivers, swamps and dams, should be selected. Such mosquito breeding sites are typically shallow depressions that are flooded during prolonged periods of rainfall and along irrigation channels.
Sentinel herd monitoring in Mali
In order to be effective and more reliable, sentinel herds should be monitored in conjunction with the monitoring of other risk indicators such as climatic parameters (see paragraphs below).
Once herds have been identified, livestock owners are informed about the background and the importance of the study and asked for their cooperation. Incentives such as free antiparasitic drugs for internal parasites should be provided during each visit to ensure the owners cooperation. If possible, no acaricides, pour-on or insecticides should be provided since their application will influence the attack rate of the animals from potential arthropod vectors.
About 30 young female sheep or goats (with two permanent incisors) are identified and permanently marked. Choosing young females reduces the probability that these animals are slaughtered or sold between visits. The animals should be at least one year old, ideally between 12 and 15 months. At the first visit, blood samples are collected from the animals and tested for IgG antibodies and IgM antibodies to the RVF virus. Only antibody negative animals should be included in the sentinel herd. If some of the animals are seropositive they should be excluded from the monitoring exercise and replaced by seronegative animals.
The sentinel herds should be visited at regular intervals. Ideally, animals should be sampled at the beginning of the rainy season and thereafter every four to six weeks up to the end of the rainy season. In a typical year this would involve four or five visits to each herd. At each visit some basic information is obtained and blood is collected from all the sentinel animals. Samples should be forwarded to the National Central Veterinary Laboratory and tested within two days of arrival for the presence of IgM and IgG antibodies.
Sentinel herd monitoring in Mauritania
Attempts should be made to keep the size of the sentinel herd above 20 animals at each location. This means that during most visits it will probably be necessary to identify new animals since some animals of the established herd may have died or have been removed for other reasons. If animals have seroconverted they will also be excluded during the next visit and replaced by new fully susceptible animals.
The principle here is to maintain at any time a basic number of animals (sheep and goats) that are well identified and fully susceptible to RVF infection at the locations prone to the emergence of RVF epidemics, and to follow these animals closely through clinical and serological investigations to detect the emergence of RVF epidemics in time.
The three essential prerequisites for an epidemic to occur are a susceptible livestock population, a massive buildup in the populations of vector mosquitoes and the presence of the RVF virus. Assuming the continuing presence or at least the close proximity of the virus in regions where the disease has occurred previously, the first two factors become the key to early forecasting of likely RVF activity.
Early work on forecasting was centred at a study site in Kenya where ground truth data for RVF virus activity had been generated for many years. Periodic outbreaks of RVF over a 40-year period were found to correlate with the positive value of a statistic based upon the number of rain days and the quantity of rainfall.
Correlation of RVF epizootics with the persistence of rainfall
The three-month rolling mean value formed a positive spike when RVF virus activity occurred and this was a function of cumulative persistent rainfall, rather than heavy precipitation over a short period. Data were based upon longitudinal rainfall data generated and recorded in the old-fashioned manner. The characteristics of the intertropical convergence zone were also important as a determinant of prevailing conditions conducive to RVF virus activity. These data allowed forecasting of RVF outbreaks with a four to ten week period during which vaccination could be carried out before cases occurred.
More sophisticated studies were possible when remote sensing satellite data (RSSD) became available. These data enabled national and regional monitoring of rainfall and climatic patterns and their effects upon the environment. Cold cloud density (CCD) measurements are closely correlated with rainfall and have replaced the laborious daily collection of rainfall data from many stations. Climatic patterns are regional in East Africa and the Horn of Africa and may be studied on this basis. A detailed analysis was made with virus isolation data over a 25-year period and the normalized differentiated vegetation index (NDVI) for the study areas. NDVI data are derived from probes measuring relative greenness and brownness of the vegetation. As the water table rises to the point where flooding may occur, the ratio approaches 0.43 to 0.45. This point was reached at each of the epizootic periods in the study period.
More recent retrospective studies using the same ground truth data have included the surface sea temperatures (SST) for the Indian and Pacific Oceans. When these were combined with NDVI data, they approached 100 percent accuracy in predicting periods of RVF virus activity during the study period. This has a pre-epizootic predictive period of two to five months before virus activity occurs.
New statistics have been derived from satellite data, known as basin excess rainfall monitoring systems (BERMS). These measure rainfall in the catchment areas of river/wadi systems and are based upon digital maps of basin and river networks. They can predict periods when flooding might occur, which is particularly valuable for the floodplain zones in the Horn of Africa countries and the Arabian Peninsula. Early data suggest that BERMS might be able to predict virus activity five months before its occurrence. The advantages of RSSD for RVF predictive epidemiology are in the relatively low costs of the systems used for analysis. These are readily available on a country and regional basis and give time for preventive measures such as the vaccination of susceptible stock and mosquito larval control methods, wherever possible.
Maps representing NDVI difference in January 1997, 1998 and 1999
International agencies are best placed to analyse satellite and other data and to provide risk countries with early warning about likely weather patterns conducive to increased RVF activity. FAO, through its Global Information and Early Warning System on Food and Agriculture (GIEWS) and the Emergency Prevention System for Transboundary Animal and Plant Pests and Diseases (EMPRES)/Livestock Programme intends to take a central role in generating these data on a continuing basis, thus providing an early warning/risk assessment service.
It must be recorded that little work has been done in other parts of Africa to validate the RSSD systems because the ground truth data have not been available and it takes many years of dedicated work to generate such data. Recent outbreaks in Somalia and northeast Kenya in 1997-98 showed retrospectively that the foci of RVF virus activity in these countries could be correlated with high NDVI values. However, more validation work is needed before the use of such techniques becomes an operational early warning system (see maps representing NDVI difference in January 1997, 1998 and 1999).
This is the most effective means to control RVF. Early warning of high-risk periods for the disease is possible and this information should drive strategic vaccination campaigns. The most effective vaccine is the modified live Smithburn neurotropic strain (SNS). This vaccine is immunogenic but has the disadvantage that it can cause foetal pathology and abortion in pregnant sheep of susceptible genotypes. Up to 30 percent of such animals may be affected by abortion or foetal abnormalities. Inactivated vaccines have been prepared but are often poorly immunogenic. Onderstepoort Biological Products in South Africa produce an inactivated vaccine that is based on a bovine virulent RVF isolate, adapted and produced in cell culture. The vaccine is then inactivated and mixed with aluminium hydroxide gel as adjuvant. It has the advantage of being suitable for use in pregnant ewes. Given the poor antibody response in cattle, the inactivated vaccine is recommended even in cows so that they can confer colostral immunity to their offspring. A booster three to six months after initial vaccination is required, followed by annual boosters.
Routine vaccination when animals are not pregnant is recommended. The SNS vaccine is perfectly safe and protective in cattle. Vaccination is NOT recommended once evidence of epizootic virus activity has been confirmed. Apart from being too late, needle propagation of the virus is a real danger.
Vaccine development. Other modified live virus and molecular derived RVF antigens are being developed, but are not currently available for field use.
The MP 12 strain was developed by mutagen induced changes in the ZH 548 strain of the RVF virus and Clone-13 is a cloned population, obtained from a field strain isolated from a mild human case in the Central African Republic. Both have been shown to be good immunogens in mice, and produce antibodies detectable by ELISA and plaque reduction neutralization assays. The Protective Dose (50 percent) (PD 50) for Clone 13 was 100.1 TCID50 and for MP 12 was 103. The S segment of the RVF virus determines virulence/loss of virulence and the NSs deletion results in attenuation. The role of NSs has been elucidated: it is an antagonist of type I interferon production. Indeed, infection of mice with strains that possess an efficient NSs does not lead to any production of interferon, whereas high levels of interferon were observed in mice infected with the NSs defective Clone 13 virus. Clone 13 is of interest because of the low risk of reversion in this virulence/attenuation marker. However, the L and M segments do not contain markers for attenuation. If animals are vaccinated when virulent strains are circulating, there is a possibility that reassortment may occur; in this case Clone 13 would induce a viraemia (which is not observed in mice). The majority of reassortants could become virulent.
An R566 strain has been derived from Clone 13 and MP 12 by reassortment in Vero cells: it contains the S segment of Clone 13 and the L and M segments of MP12, which contain seven and nine point mutations compared with their virulent parent. Some of them induce attenuation and thermosensitivity. Thus R566 is safe, because of its attenuation in the three segments of the genome. R566 has been shown to protect mice in the laboratory.
Strategic larvicidal treatment of mosquito breeding habitats is recommended. Both hormonal inhibitors such as methoprene and larvical toxins such as those produced by Bacillus thurigiensis give excellent results and both are commercially available. However, they may be difficult to use in some places with floods of wide distribution. The widespread use of vehicle or aerial mounted ultra low volume insecticide sprays appears to have limited effect upon RVF transmission rates or the target adult mosquito species.
These do not appear to have any effect upon the course of an outbreak within an infected country. They may, however, be relevant to the movement of animals for trade from enzootic/epizootic areas, where RVF virus transmission is occurring. In this situation, viraemic animals could arrive in an uninfected country within the incubation period for the disease. If this should happen and there were large numbers of mosquito vectors present, capable of RVF virus transmission, then the possibility of introduction of RVF is very real. For this reason, all export of livestock should be banned during RVF epizootic periods.