Seroepidemiology of Crimean-Congo Haemorrhagic Fever among cattle in Cameroon: Implications from a One Health perspective

Background Crimean-Congo Haemorrhagic Fever (CCHF) is a tick-borne viral zoonotic disease distributed across several continents and recognized as an ongoing health threat. In humans, the infection can progress to a severe disease with high fatality, raising public health concerns due to the limited prophylactic and therapeutic options available. Animal species, clinically unaffected by the virus, serve as viral reservoirs and amplifier hosts, and can be a valuable tool for surveillance. Little is known about the occurrence and prevalence of Crimean-Congo Haemorrhagic Fever Virus (CCHFV) in Cameroon. Knowledge on CCHFV exposure and the factors associated with its presence in sentinel species are a valuable resource to better understand transmission dynamics and assess local risks for zoonotic disease emergence. Methods and findings We conducted a CCHFV serological survey and risk factor analysis for animal level seropositivity in pastoral and dairy cattle in the North West Region (NWR) and the Vina Division (VD) of the Adamawa Region in Cameroon. Seroprevalence estimates were adjusted for sampling design-effects and test performance. In addition, explanatory multivariable logistic regression mixed-effects models were fit to estimate the effect of animal characteristics, husbandry practices, risk contacts and ecological features on the serological status of pastoral cattle. The overall seroprevalence was 56.0% (95% CI 53.5–58.6) and 6.7% (95% CI 2.6–16.1) among pastoral and dairy cattle, respectively. Animals going on transhumance had twice the odds of being seropositive (OR 2.0, 95% CI 1.1–3.8), indicating that animal movements could be implicated in disease expansion. From an ecological perspective, absolute humidity (OR 0.6, 95% CI 0.4–0.9) and shrub density (OR 2.1, 95% CI 1.4–3.2) were associated with seropositivity, which suggests an underlying viral dynamic connecting vertebrate host and ticks in a complex transmission network. Conclusions This study demonstrated high seroprevalence levels of CCHFV antibodies in cattle in Cameroon indicating a potential risk to human populations. However, current understanding of the underlying dynamics of CCHFV locally and the real risk for human populations is incomplete. Further studies designed using a One Health approach are required to improve local knowledge of the disease, host interactions and environmental risk factors. This information is crucial to better project the risks for human populations located in CCHFV-suitable ecological niches. 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). 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. The Roslin Institute, at the Royal (Dick) School of Veterinary Sciences, University of Edinburgh, UK, approved the serological assessment of CCHFV antibodies in the serum bank in 2019. Cameroon is an ecologically diverse country in Central Africa covering an area of 475,440 km2. It borders Nigeria, the Republic of the Congo, Gabon, Equatorial Guinea, Central African Republic, and Chad as well as a coast on the Gulf of Guinea. Climate dynamics can be generally thought of in terms of a wet (May–October) and a dry period (November–April) with rainfall and temperature varying monthly according to the season [40,41]. The country is organized in 10 administrative Regions; these Regions are further split into Divisions and sub-Divisions [42]. Major cattle producing areas are the North West, North, Extreme North, northern parts of the East and Central, and the Adamawa Regions. Overall, the cattle population has increased over recent years and the country is a recognized livestock producer in the Central-West African region [40,43]. The current analysis is focused on data collected from animals reared in the North West Region (NWR) and Vina Division (VD) of the Adamawa Region (Fig 1). The red area shows the North West Region and its Divisions. The blue area shows the Vina Division of the Adamawa Region and its sub-Divisions. Shapefile obtained from GADM database, freely available for academic uses with permission from Global Administrative Areas (https://gadm.org/maps/CMR.html). The figure was made with RStudio version 3.5.3. The Adamawa Region occupies a 64,000 km2 territory, generally located over 1,200 m and classified as Guinea savannah, characterized by woodland and grass savannah vegetation [41,44]. Rearing cattle is the main economic activity of the region and it is focused on pastoralist systems primarily managed by local ethnic groups; however, some residents are crop growers working under the principle of a collaborative agricultural economy [44]. Within the Adamawa Region, the VD has a land area of 17,196 km2 with altitudes ranging between 500–2,500 m; the area is topographically characterized by a mountainous western border that softens as it reaches east into an undulating grassland savannah [40]. Similarly, the NWR occupies an area of 17,300 km2 with distinctive rocky-mountains rising between 700 and 3,000 m, although subtropical forest and plateaux savannah are also present [40]. In terms of economy, the agricultural sector is strong and represents the main source of income in the rural areas (~ 80%). Furthermore, it is estimated that 60% of the NWR is a viable terrain for livestock production leading to an active involvement in beef farming [45]. Fulbé, Mbororo, Niam Niam, Laka, Mboum and Baya ethnic groups populate the North West and Adamawa Regions. Some of them (e.g. Fulbé and Mbororo), are part of the Fulani ethnic group, widely extended across sub-Saharan West and Central Africa and recognized as the main pastoral community [40]. In 2013, the cattle populations of the NWR and VD were estimated to be 546,508 and 176,257 respectively, with herd sizes ranging from 50 to 150 cattle [40]. The predominant breed is the Bos indicus Fulani cattle but other improved breeds such as the Gudali and crossbreeds are also widespread particularly in the Adamawa Region. The latter breeds provide either better productivity or resilience against the harsh conditions of the territory [40]. Cattle are normally grazed in an extensive system on communal pastures close to the farm [40,45]. During the dry season, transhumance, a pastoral practice involving long-distance movement of animals takes place, with the aim of overcoming the seasonal shortages in pasture availability [46]. Intensive farming systems for cattle production are not a common practice. However, semi-intensive farming, primarily based on a cut-and-carry feeding system, is of growing importance for the dairy sector where imported Holstein-Friesian crosses are used. But this remains a very small proportion of the livestock industry in the country [45]. Pastoral herds located across the NWR and VD and dairy herds from the NWR were studied to estimate CCHFV antibody prevalence based on serum bank samples available from two previous cross-sectional studies investigating the epidemiology and phylogenetics of bovine tuberculosis and liver fluke infections in Cameroon [40]. Pastoral cattle located at the NWR and VD comprised the main sampling frame, which was built based on the official vaccination records for 2012 (Sampling frame: 5,053 pastoralists’ herds). Conversely, dairy herds from the NWR were retrieved and sampled based on data from the three largest dairy cooperatives registered for the area in 2012 (Sampling frame: 164 dairy herds). A herd was defined as an established group of animals managed collectively as a unit under a well-structured ownership model. In both cases, the Ministry of Livestock, Fisheries and Animal Industries (MINEPIA) provided the records as the closest representation of the true number of herds according to an official source. Sampling and data collection took place from January to May 2013 in the NWR and from September to November 2013 in the VD. The list of pastoralist herds in each site was stratified by administrative area; seven Divisions in the NWR and eight sub-Divisions within the VD. A random sample of 50 herds was taken proportional to the total number of herds listed per study site in the NWR and the VD. In each herd, 14–15 animals were selected by quasi-random sampling, stratified to three age classes: 6 months– 2 years-old, 2–5 years-old and >5 years-old, termed young, adult, and old groups, respectively. Likewise, a stratified random sample of 46 small-scale dairy herds was selected proportional to the number of dairy herds per cooperative in the NWR; all animals were sampled per herd (one to three animals per smallholder). Herd replacement by resampling was applied in both scenarios if herdsmen were unwilling to engage with the study or when unforeseen logistical situations prevented visiting one of the originally selected herds. In-depth information about study design, sample size calculations and sampling methods used in the original study has been reported by Kelly et al. (2016) [47]. Biobanked cattle serum samples were available for processing at the Roslin Institute at the Royal (Dick) School of Veterinary Medicine (University of Edinburgh), UK. Before transportation to UK, all samples underwent water-bath heat treatment for 120 minutes at 56°C. Serum samples were labelled to allow linking back to herd and animal level questionnaire data and stored at -20°C until processed. Individual and herd level data was available from a structured questionnaire administered to each herdsman in Fulfulde. The questionnaire covered aspects of cattle husbandry and management, dairy practices, individual animal features and GPS location of the farm. A copy of the questionnaire and further details on data collection are reported by Kelly et al. (2016, 2021) [47,48]. Age was estimated by dentition score according to the number of permanent incisor teeth present and classified from < 2 years (no permanent incisors) to ≥ 5 years (all incisors in wear/broken). Calves defined as less than 6 months old by the herdsman were excluded from the sampling to avoid potential issues of maternal antibodies. Breed was defined according to phenotypic traits as either mixed breed (mixed Bos indicus), exotic (any pure Bos taurus or crossbreeds Bos indicus x Bos taurus), Fulani and Gudali [40]. While Fulani and Gudali are considered local breeds, the Gudali cattle has been systematically improved through a government-led programme over the last 50 years [49]; therefore, Fulani was considered the indigenous breed for the study area. Laboratory analysis was carried out in February and March 2019 at the facilities of the Roslin Institute at the Royal (Dick) School of Veterinary Medicine (University of Edinburgh), UK. Anti- CCHFV Immunoglobulin G screening (IgG) was performed by a commercial double-antigen Enzyme Linked Immunoabsorbant Assay (ELISA) (ID.vet, Grabels, France) based on the recombinant nucleoprotein antigen (NP), with a reported diagnostic sensitivity of 98.9% (95% CI 96.8–99.8) and specificity of 100% (95% CI 99.8–100) [50]. The test was performed in 96-well microplates following the manufacturer´s instructions [51]. Duplicate kit positive and negative controls were included per plate. Optical density (OD) values were determined spectrophotometrically using an automated ELISA reader (Themocientific Multiskan Go) set at 450 nm. The validity of the test results was verified for every plate based on the manufacturers recommendations of an OD for the positive controls > 0.35 and a ratio of the mean OD values of the negative and positive controls > 3. The calculation of the S/P% as the ratio of the sample OD and the mean OD of the positive controls of the plate was used as the output measure and an S/P percentage >30% were recorded as positive. Quality control was performed for a total of 180 samples (2 plates), with 100% concordant results. Individual S/P% values were collected per processed plate by using a standard report datasheet and combined into a final database. All data analyses were performed using R packages and functions in RStudio version 3.5.3 [52,53]. Figures, graphs and maps were plotted using the ggplot2 package [54]. The shapefiles of the country maps and its administrative divisions were obtained from the open access database of Global Administrative Areas (GADM) [55]. A design effect correction, accounting for the stratified population structure of the pastoral cross-sectional study was implemented [56]. Clustering (herd identification), strata information (Divisions and sub-Divisions) and animal and herd sampling weights were combined into a nested complex survey object using the svydesign function in the Survey package [56]. This survey object was then used within the package’s summary functions to obtain CCHFV survey design-adjusted seroprevalence estimates. Thereafter, seroprevalence values in pastoralist and dairy herds were corrected for test performance using the Wilson’s method to provide appropriate confidence intervals for the adjusted seroprevalence, while accounting for the imperfect test sensitivity and specificity [57]. Within the pastoralist subset, two separate explanatory multivariable mixed-effect logistic regression models were used to estimate the effect of individual animal features, risk contacts, animal husbandry practices and environmental variables on CCHFV serological status of cattle. Both herd and administrative Division or sub-Division were included as random terms to account for the clustering effect of the sampling design. Global models were built using all the selected fixed effects through the glmer function in the lme4 package [58]. The first model explored the effect of individual traits, risk contacts and animal husbandry practices (“Individual risk-factor model”). Preliminary variable selection considered central biological or epidemiological attributes related to disease risk as per its potential causal dependency network. Statistically significant variables based on univariable analysis with a cut-off value of p≤0.2 were also included [59]. A multi-correlation matrix was used to check for the presence of highly correlated explanatory variables. Multi-model inference was performed in order to reduce variable selection bias, achieve a better precision and approach model selection uncertainty. Model averaging was used to estimate the final coefficients using the MuMIN package [60]. A subset of models was generated based on all the potential combinations of the fixed effects considered at the global model; each candidate model was assessed by the delta Akaike´s Information Criterion (AIC). Candidate models with a delta AIC (Δi) ≤ 2 were averaged, as they are considered to be as good as the best model [61]. The second model focused on the role of ecological covariates (“Ecological model”) on CCHFV seropositivity by means of a standard logistic regression model adjusting for the influence of significant features identified through the Individual risk-factor model. Animals that went on transhumance were purposively removed from the analysis to reduce the possibility of bias introduced by animal movement on the inference of ecological features for past disease exposure. Elevation at the georeferenced point was extracted from an SRTM30 digital elevation model (DEM). Modelled climate data was downloaded from the UEA Climate Research Unit (version 4.03)[62]. The mean of the mean, minimum, maximum temperature, vapour pressure (used to calculate absolute and relative humidity), and precipitation for the years 2011–2014 were extracted at the location of each georeferenced point. To test the immediate weather impacts on seropositivity, we compared the climate average of these to the mean values during a 90-day window prior to sampling. Landcover variables were downloaded from the European Space Agency (ESA) climate change initiative (CCI) landcover classification [63]. To describe the landcover in the area surrounding the farm, the number of pixels of grassland, shrubland and trees within 5 km of each point were extracted. Climate and spatial variables were rescaled to aid model fitting as required. Model selection was performed through the AIC criterion. Final model estimates were converted from the log scale to OR and presented in conjunction with its 95% CI. Model diagnosis was conducted through visual inspection of plots and goodness-of-fit measures for hierarchical regression models with a binary outcome; residual diagnosis, coefficient of determination (D) and ROC curves were assessed [64–66]. In addition, the Intraclass Correlation Coefficient (ICC) was calculated to indicate the proportion of variance explained by the clustered study population [67]. Details on the procedures used for model selection and assessment of model performance and fit are documented in S1 Appendix (“Individual risk-factor model”) and S2 Appendix (“Ecological model”).