Rift Valley fever (RVF) is a zoonotic viral disease of ruminants, caused by a Phlebovirus in the Bunyaviridae family transmitted by floodwater mosquitos. It was first described in Kenya in 1931, and has since been reported in many countries in East and Southern Africa, as well as the Arabian Peninsula and sporadically in West Africa (1–3). It is considered an important emerging zoonotic pathogen of public health significance as its range is expanding, which is linked to climate change and expansion of the vector habitat. RVF affects mainly African communities with low resilience to economic and environmental challenges (4). Its epidemiology is characterized by explosive epidemics in both human and livestock populations, where livestock abortions can be a useful sentinel event for human disease risk (5). These events are often associated with flooding such as in El Niño years or dam construction and are followed by long inter-epidemic periods where there is little evidence of the viral presence in those populations affected by epidemics (6). RVF virus (RVFV) epidemiology of persistence during inter-epidemic periods as well as in areas without apparent epidemics is still poorly understood (4).

Aedes and Culex mosquitoes are the main vectors and potential reservoirs of the RVF virus (RVFV). The virus can be transferred vertically from female mosquitoes to their eggs in some species of the Aedes genera (7–9). Sheep, goats, and cattle are the domestic species most likely to be clinically affected but signs are usually mild or inapparent in adult animals. However, major outbreaks of abortions and death in neonates can occur during epidemic periods which result in significant direct economic losses (9–11). The virus can also infect wildlife species including Cape buffalo, as well as spillover into humans (12). The RVFV is transmitted between animals and humans through the bite of an infected mosquito vector. The disease in humans can also result from direct contact with infected tissues, blood, or body fluids (13). A rise in RVFV prevalence in domestic ruminants can sometimes precede epidemics in humans (5) and similarly a decline in herd immunity in the inter-epidemic periods coupled with extensive flooding appears to facilitate these explosive outbreaks (6). Symptoms of the disease in humans can vary, ranging from flu-like symptoms to more severe conditions such as meningoencephalitis, hemorrhagic fever, or death (9, 13). The case fatality rate for patients developing the hemorrhagic form of the disease can be as high as 50% (4).

Epidemiological studies of RVF have mostly focused on East Africa (14), where the virus was first isolated, with less information about its significance in Central and West Africa. However, outbreaks in human populations in Mauritania and Senegal have been associated with the development of new dam projects (4). In Central Africa, livestock exposure to RVF has been reported in the savanna of northern Cameroon [3–20% within small ruminant herds (2, 15, 16)] and Chad [4.4% in cattle, 10.7% in sheep, and 8.6% in goats (17)]. Most recently, in a large sample across Cameroon, seroprevalence estimates of 13.5% (11.4–15.7) for cattle and 3.4% (2.3–4.7) for small ruminants were reported (18).

Cameroon is a significant cattle producer in the Central-African region, with livestock contributing ~$476 million to the national economy in 2010 (19) and being of cultural importance to rural communities, particularly the pastoralist Fulani communities. The Northwest Region (NWR) and the Vina Division (VD) of the Adamawa Region of Cameroon are major cattle-keeping areas in the wider Adamawa Plateau of Central Africa. Cattle are kept for many reasons, including financial, draught power, dairy products, and trade. The area is mostly covered by sparse tree savannah, with a dry season between November and April, and a wet season from May until October (20). Culex spp. and Aedes spp. mosquitos are present in Cameroon (21) and due to the close association between cattle and people in the country, cattle may act as a reservoir for RVFV.

Despite the favorable climate for RVFV vectors and the abundance of livestock, little is known about the epidemiology of RVFV in Cameroon (18). This study aims to get population-based estimates of the seroprevalence of RVFV antibodies in cattle populations in two major cattle rearing regions of Cameroon and identify management or environmental factors associated with increased risk of individual animal-level seropositivity. In addition, we aim to estimate the force of infection and basic reproductive number.

The authors have followed the STROBE (22) and recent STROBE-vet (23) guidelines for reporting observational epidemiological studies.

A total of 1,498 cattle were sampled from 50 herds in each of the two study sites in Cameroon. The raw percentage positivity (PP) from the ELISA readings is plotted in Figure 2A and suggests a bimodal distribution. The majority (n = 1,381) of the animals were classed as seronegative with a distribution of PPs above the 40% manufacturers cut-off and a much smaller distribution of samples below 40% classed as positive. The individual binary (positive/negative) results are further plotted in Figure 2B. The vertical axis is age in years and the horizontal axis is herd rank-ordered by within-herd seroprevalence and stratified by the study site. This shows that more of the higher seroprevalence herds are in the VIN, although the highest six seroprevalence herds were in the NWR. The higher seroprevalence herds also have positives scattered across the ages.

The proportion of seropositive herds (i.e., herds with >0 positive animals) was 42% (95% CI: 28.5–56.7) for the NWR and 68% (95% CI: 53.2–80.0) for the VIN and was statistically significantly different between the two sites (two-sample test for equality of proportions, p-value = 0.009). The animal-level seroprevalences per division/sub-division, adjusted for the sampling design, are presented in Table 1. The overall design adjusted animal-level seroprevalence of RVF for the NWR was 6.5% (95% CI: 3.9–11.0) and the VIN was 8.2% (95% CI: 6.2–11.0). Although the overall seroprevalences at the two sites were largely similar, there was considerable variation between the different administrative strata as can be seen in Table 1 and Figure 3. Ngoketunjia in the NWR and Mbé and Martap in the VIN had particularly high seroprevalences. Ngoketunjia includes a large dam project at its southern end but there are no obvious large water bodies that could simplistically explain higher seroprevalences in Mbé and Martap.

The age-stratified seroprevalence and the force of infection stratified by the site are plotted in Figure 4. As would be expected for an infectious disease like RVF, seroprevalence increases with age (and thus the time at risk of infection) and assumes that animals will remain seropositive after exposure. In both populations, the predicted seroprevalence does not appear to go above ~20%. The force of infection was modeled using three different models as shown in the various red lines in Figure 4. Their AICs are given in Table 2 and do not suggest that the more complex models are a markedly better fit for the data. The linear and quadratic models were the best for the VIN and NMR, respectively. The NWR model may be overly influenced by a small sample of older animals that had a particularly high seroprevalence but age-independent FOI is around ~0.029 for the VIN and ~0.024 for the NWR. Whichever model one chooses, the results suggest generally low FOI possibly declining with age. The FOI for VIN was slightly larger than for the NWR which may be related to the higher number of positive herds at this site. The effective reproductive rate R_t for the VIN was 1.10 compared to the NWR, where it was estimated as 1.08 suggesting little difference from a practical point of view.

A mixed-effects multivariable logistic regression model was developed to explore the association of various risk factors with seropositivity at the animal-level adjusting for clustering by the herd. The model-building strategy is given in Table 3. The odds ratios for the final model are presented in Figure 5. The risk of being seropositive increased with age (as would be expected) and also with contact with antelopes (though the exact species were not known by the herdsmen). In contrast, the odds of seropositivity decreased with higher altitude and higher temperatures. It also decreased with contact with buffalo suggesting a complex relationship with wildlife contacts. Finally, the administrative strata were included as a fixed effect, but there is little supporting statistical evidence that there is a residual strata difference. The intra-cluster correlation is relatively small at 0.13 suggesting that most of the variance is at the animal rather than the herd level. The marginal R² (the variance explained by the fixed effects) is ~32% and the conditional R² (the variance explained by both the fixed and random effects) is ~41%, with an AUC of ~78% suggesting the model is useful and explains a reasonable amount of the variation.

This study is the first population-based livestock study of RVF in Cameroon to estimate the effective reproductive rate R_t, the FOI, and identify risk factors for RVF seropositivity in cattle in the country. It highlights that there is a high seroprevalence in the cattle population and therefore may be an important differential for abortion in cattle (and other small ruminants or camels). The estimates for the VIN, 68% (95% CI: 53.2–80.0), and NWR, 42% (95% CI: 28.5–56.7), are similar to those reported by Rissmann et al. (18) but are not directly comparable as their estimates do not separate species or adjust for population sizes and sampling. However, that RVF has now been reported in small ruminants and cattle by several studies also emphasizes its importance as a zoonoses, particularly, for herdsmen, women and children, and abattoir workers. A meta-analysis by Nicholas et al. (36) identified several specific human risk factors including being male, handling aborted animal tissues, helping with birthing, skinning an animal, slaughtering an animal, and drinking raw milk. These are all activities many herdsmen and slaughterhouse workers as well as wider family members of herdsmen are likely to be carrying out and therefore RVF needs to be on the differential lists of clinicians. In addition, the infected cattle population will be acting as a major reservoir of infection for the region. Interestingly, RVF is not widely reported in humans in Cameroon and they do not appear to currently suffer the large abortion storms in ruminants seen in East and North Africa. This may in part be explained by the lack of reporting due to the remote nature of the setting and lack of access to health or veterinary care. It may also suggest that there is something different about its epidemiology in the region or may reflect the lack of diagnostic effort or both. Anecdotal evidence from discussions with local clinicians highlighted that a high proportion of cases in hospitals are fevers of unknown origin that tend to be undiagnosed. A recent small study of African rainforest hunter-gatherers, known as the Baka ethnic group, in Eastern Cameroon reported a 12.4% seropositivity in this population (37). However, there needs to be much more comprehensive surveillance and diagnostic capability in Cameroon and West and Central Africa, generally so that both veterinary and human clinical cases can be identified and correctly managed.

Infection appears to be widespread across both the NWR and VIN with some evidence of hotspots such as Ngoketunjia, Mbé, and Martap. The overall seroprevalence appears to be <10% in these two populations compared to high clinical incidence areas in South Africa where they report 42% seropositivity in cattle (38). The age-independent FOI (the probability an individual becomes infected in a given year) estimates of ~0.029 for the VIN and ~0.024 for the NWR are in line with other estimates in cattle in Madagascar (39) and give a useful starting value for future modeling. However, the best fitting models suggest a more complex relationship with FOI declining slightly with age.

Entomological risk factors include temperature, rainfall, and biotic factors such as breeding sites and vertebrate hosts. Cameroon has plenty of potential vertebrate hosts and high seasonal rainfall in the areas studied, although typically you do not tend to see the flooded pans described in East and Southern Africa. There is little in the literature about the species competence of mosquitoes in Cameroon to transmit RVF. Simonet et al. (21) looked at Northern Cameroon and identified 9 primary and 22 secondary vector species capable of transmitting RVFV, and the authors highlight the need for PCR analyses of potential vectors to understand their true importance.

The role of wildlife may also be important. Here, we observed a very strong association with increased risk when herds reported contact with antelope, but a reduced risk when they reported contact with buffalo. Only antelope may be a reservoir in these areas and buffalo are currently not infected. Several studies have reported high seroprevalences in wildlife species including various antelopes and buffalo but their importance is less clear from an epidemiological point of view (40). In this Kenyan study, seroprevalence was especially high in some wildlife species. There may be vector-feeding preferences linked to species of wildlife and how they interact with livestock that we currently do not understand. In Cameroon, there are increasing conflicts between pastoralists and crop farmers (41), and this is changing patterns of grazing and transhumance that will impact transmission potential with wildlife.

Much of the variation is explained by altitude, with lower altitude areas having a higher risk, but this is countered by a reduced risk at higher temperatures suggesting a complex environmental interaction presumably driving mosquito breeding and feeding habits and virus replication and amplification within the vector. The age-stratified analysis suggests a more stable endemic epidemiological pattern rather than a more periodic epidemic outbreaks pattern. Although correlation between RVF outbreaks and the warm phase of El Niño/Southern Oscillation (ENSO) phenomena which lead to abnormal rainfall has been reported (7), there have been instances where no outbreaks were recorded following seasons of exceptionally above normal rainfall (42). Moreover, in some sub-Saharan regions, such as West Africa, RVF outbreaks are not known to be correlated with above-average rainfall (43). The lack of reports of either livestock or human outbreaks and the R_t of ~1.08 as well as the clear pattern of increasing risk of exposure with age may mean that there is a different epidemiological process and a more stable low impact endemic situation in Cameroon possibly related to a different climatic cycle less influenced by El Niño. This study has focused on cattle because a sera biobank was available for screening. However, given the high seroprevalence of RVF as well as other zoonoses such as Congo-Crimea hemorrhagic fever (44), brucellosis leptospirosis, and Q fever (45), there is an urgent need for a more One Health-based approach looking across species as well as human populations to understand the real impacts of these diseases.

RVF virus appears to be circulating widely in the livestock rearing areas of Cameroon. However, there are no reports of livestock abortion storms or human clinical disease, though this may be due to limited investigations conducted in the country. The serological results suggest a more endemic epidemiological pattern that may be different from the cyclical epidemics seen in East Africa linked to El Niño. There is an urgent need for One Health-based study to understand the clinical scale and impacts of the RVF virus both on the livestock and human populations in Cameroon to direct interventions to reduce disease burden and antimicrobial misuse.

An anonymised version of the data included in the article is available on request to the corresponding author.

The animal study was reviewed and approved by the Institute of Research and Development (Cameroon) and the University of Edinburgh Ethics Committee (UK) approved the study at the moment of data collection (VERC No: OS02-13). Written informed consent for participation was not obtained from the owners because verbal permission was obtained from all herdsmen in order to collect the biological samples from the animals and before administering the questionnaire. A brief explanation of the purpose and procedures of the study preceded the consent and herdsmen were informed of the possibility of opting out at any stage. Verbal consent was used as most herdsmen had limited formal education.

BB, RK, KM, VNT, and LN conceived and designed the project. RK collected the data. RK, JB, and VNN processed samples and conducted ELISA tests. EF, RC, SM, and BB conducted the data cleaning and analysis. BB, EF, and JB drafted the manuscripts. All authors contributed to the article and approved the submitted version.

BB acknowledges the support provided by the Wellcome Trust [WT094945]; this grant funded the original fieldwork. BB is supported by BBSRC Institutional funding (BBS/E/D/30002275). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Importantly we would like to thank all the cattle keepers, MINEPIA veterinary centre staff and delegates whom without this study would not possible.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.