Rabies in Uganda: rabies knowledge, attitude and practice and molecular characterization of circulating virus strains

A total of 35 brain tissues were collected from suspected dogs, cattle, goats, sheep, fox, jackals in seven districts of Uganda (Moyo, Ntoroko, Jinja, Kabaale, Kabongo, Namayingo and Kabarole districts) (Fig. 1).

The survey was conducted in the two districts of Moyo and Ntoroko to collect data based on dog biting history reported by the selected households. These two districts were selected for the KAP study because of their high reporting of rabies suspected samples to NADDEC, for laboratory rabies confirmation, as compared to any other districts in Uganda from 2009 and 2013. Eighty-four (Moyo, n = 44; Ntoroko, n = 40) households affected by dog bites participated in the study, as well as the local government health centers and medical private centers where first aid treatment due to dog bites were offered. Data was collected using a well-designed tool with both objective and open-ended questions (Supplementary Material: SM0 - questionnaire). Collected parameters include knowledge status of the households about rabies, handling of dog bites by local community, age, dog bite site, death associated with suspected rabies and, affected livestock. Dogs were defined as owned or confined, free roaming or stray dogs in the community.

An animal bite was defined as any injury or wound inflicted by a suspected rabid animal to humans and registered during the review period. In 2013, patient and availability of anti-rabies vaccine, PEP protocols in the selected health facilities were reviewed. Data on age, gender, number of patients bitten by suspected rabid dogs and other animal species involved in biting people were also recorded. The survey also focused on challenges in the management of dog bite victims in the visited medical centers.

From 2009 to 2013, the access to the livestock disease register provided the number of dogs vaccinated, target areas, suspected number of rabid animals, animal species involved in biting humans and health status of animals. Methods of controlling stray dogs if any were recorded as well as the number of livestock, wildlife and dogs suspected of rabies, all data provided by the district veterinary officer.

Data were registered using Microsoft Excel 2007 (Microsoft, Seattle, WA, USA), and analyzed using STATA 10 package for statistical analysis [9]. The Fisher’s exact test or the Chi-square X² test was applied according to the number of observations with a statistically significant threshold set at a P < 0.05 (95% confidence interval).

A number of categorical variables were stratified (2 by 2 tables) according to the district of origin in order to assess the significance among the tested variables of interest in Moyo and Ntoroko. Differences in frequency of observations between districts and the association between key variables of interest using uni-variable analysis were calculated.

The statistical analysis performed was descriptive, qualitative and quantitative among the 84 households of Moyo and Ntoroko districts. The data captured included: age, gender, knowledge about rabies (transmission, clinical signs and management of dog bites), vaccination and treatment, ownership, care and number of bites. The reason for using the non-probabilistic method of sampling was to explore the knowledge, attitude and practice about rabies in the specifically affected households.

The brain samples were collected from dead animals which where owned by private individuals. Some were from farmers and fishermen on along river Nile. Some were from dogs that were straying in these communities. The affected goats, cattle, and sheep brain samples used in this study were willingly provided by farmers for laboratory testing, after suspected clinical signs of rabies disease were observed. Verbal consent to collect samples of dead Jackals that were found lying dead in Kidepo National Game park was given by wildlife authority, because in Uganda only National Animal Disease Diagnostics and Epidemiology Centre (NADDEC) has the capacity to diagnose rabies using the direct fluorescent antibody test method. Thirty-five (35) brain tissue samples sourced from dogs, cattle, goats, sheep, foxes and jackals were tested at NADDEC for RABV antigen detection using Fluorescent Antibody Test (FAT) as previously described by OIE in 2013 [8]. Brain specimen were stored in 50% glycerol-saline at − 80°c and later shipped to Instituto Zooprofilattico Sperimentale delle Venezie (IZSVe, Italia) for RABV culture and viral RNA extraction, and one-step RT-PCR and sequencing.

Under class II Biosafety cabinet, 30 mg of each sample was homogenized with motor and pistol to mechanically lyse the nerve cells, then 3.5 μL of cell suspension on RAI PBS, guanidinium and ß-mercaptoethanol were lysed using a vortex mixer, followed by one-minute centrifugation (12,000). The entire procedure of RNA extraction was performed by using the Nucleospin RNA II kit, following the manufacturer’s instructions (Macherey–Nagel GmbH & Co., Düren, Germany). One hundred μl of supernatant was pipetted and placed in new sterile RNA collection tubes with columns to be washed with buffer and centrifuged 1 min (12,000 rpm). After DNA digestion on silicate membrane and centrifugation, RNA was eluted from the membrane. The purified RNA was eluted in a final volume of 40 μl and subjected for amplification by One Step RT-PCR.

RT-PCR was performed using the One-Step RT-PCR kit (Qiagen®, Germany), following the manufacturer’s instructions. Published primers were used to amplify a 603 bp region of the N gene as previously described [10]. This region of the N gene is highly conserved and involved in the nucleoprotein region (450–451 amino acids) that protects the genome from nucleases and mediates evasion or induction of type 1 interferon. PCR products were visualised on 7% acrylamide gel electrophoresis (Fig. 2).

PCR products were purified with ExoSAP-IT (USB Corporation, Cleveland, OH) and sequenced in both directions using the Big Dye Terminator v3.1 cycle sequencing kit (Applied Biosystems, Foster City, CA), the Performa DTR Ultra 96-well kit (Edge Biosystems, Gaithersburg, MD) and a 16-capillary ABI PRISM 3130xl Genetic Analyzer (Applied Biosystems, Foster City, CA, USA). Nucleotide sequences were then assembled using SeqScape v.2.5 program (Applied Biosystems 78,744 Austin, Tx).

Phylogenetic analysis was based on partial N gene sequences. Original sequences were trimmed and aligned with a set of sequences representing the genetic diversity of RABV using MAFFT [11]. First phylogenetic analyses were performed in maximum likelihood using PhyML, implemented in Seaview [12]. Then, Bayesian phylogenetic analyses were implemented in BEAST (version 1.8.4) [13], using the most appropriate model according to the corrected Akaike Information Criterion (AICc) from Jmodeltest2 [14]. The model included the general time reversible model of substitution with a gamma distribution and a proportion of invariant sites (GTR + I + G). The random starting tree used the coalescent (constant size) model. A relaxed molecular clock with an uncorrelated lognormal distribution was implemented in combination with a tip date sampling [15]. The MCMC (Markov-Chain Monte Carlo) was launched 3 times with 10 E7 iterations and 10 E3 samples each in order to reach Effective Sampling Size (ESS) values above 200 after logs and trees files were combined using log Combiner with a 10% discard. In order to evaluate spacial diffusion dynamics of rabies in Africa, a discreet geographic model (i.e. based on country and associated biogeographic zone) were coded as discrete traits for each genetic sequence and incorporated into the model as an additional data partition. Symmetric substitution model and a social network using BSSVS (Bayesian Stochastic Search Variable Selection) procedure were used to infer a discrete ancestral state reconstruction and to estimate most probable geographical diffusion pathways [16]. This discrete model was complemented by a model in continuous space, by using geographic coordinates of taxa with Brownian random walk used as continuous trait model with bivariate trait representing latitude and longitude values. The ancestral state reconstruction (geographic coordinates of nodes) was performed in continuous space and for each ancestor to evaluate the most probable diffusion route of lineages. The maximum clade credibility tree with continuous traits was then combined with GeoJSON map file (https://​geojson-maps.​ash.​ms) to produce the graphical output using spreaD3 software version 0.9.7 [16].

In this study both Ntoroko and Moyo districts interviewed population equally presented their dogs for vaccination against rabies. In Moyo 38.64% of respondents owned dogs as compared to 57.50% in Ntoroko. Among the interviewed households in Moyo and in Ntoroko, 45.45 and 47.50% had access to veterinary services, respectively. Also, 36.36 and 42.50% of the respondents who owned dogs presented their dogs for parental vaccination against rabies in Moyo and Ntoroko respectively. The world rabies day (WRD) events participation represented 13.64 and 2.5% of respondents in Moyo and Ntoroko, respectively.

The association between knowing rabies (presentation, risk, transmission) and knowledge of WRD showed that all respondents had some knowledge about rabies while 10% only truly knew about rabies disease. Among the respondents who took their dogs for vaccination, 20.00% had some knowledge about rabies and 40.51% were knowledgeable about rabies. In both districts over 90% all respondents were aware about rabies as a disease of dogs, however only 18.99% of the responds washed their wounds before seeking medical treatment to nearest regional hospital.

The most affected age range of humans with dog bites episodes was between 11 and 20 years and female accounted for most cases with 51,35% and 63,89 in Moyo and Ntoroko, respectively (Table 1).

The study revealed that respondents in the two districts of Ntoroko and Moyo were knowledgeable about rabies and in agreement (P = 0.665). From these 84 respondents in Ntoroko and Moyo districts, majority of the interviewed households knew the clinical signs and how the virus is transmitted through dog bite (Table 2). In Moyo and Ntoroko, respectively 63 and 55% of the dog-bite victims that visited the hospital, were not aware of wound cleansing and report to either hospital or to the village health centres without cleaning the wound but visited the health units for PEP.

Only 36.36 and 42.50% took their dogs to the community vaccination centres in Moyo and in Ntoroko, respectively. Farmers where not attentive presenting their dogs for vaccination even some of them reported to be bitten by stray dogs. In several instances, stray dogs were then killed by shooting, chemical or poisoning and others infected stray dogs died from rabid disease.

In Moyo only 25.00% young individuals attempted to answer the questionnaires, while in Ntoroko there were 30%. In the adult age-class 43.00% respondents were from Moyo and 37.50% from Ntoroko. In the highest age-class, 27.27% respondents were from Moyo and 27.50% from Ntoroko district. In Moyo, 70.45% of respondents were male and 62.50% in Ntoroko district, while 29.55 and 37.50% were female, respectively.

Among thirty-five brain samples tested, 77.1% were positive for rabies by FAT (Table 3), and were unequally distributed among the different species (Table 4). The phylogenetic analyses showed that the RABVs detected in the study belonged to the cosmopolitan lineage, with both Africa 1A and Africa 1B clades detected (Figs. 1, 2, 3). Four distinct sub-clades (sub-lineages) of RABV circulating contemporaneously in Uganda were identified (Fig. 3b). Four sub-clades belong to the African 1B lineage and another (6 sequences) to the African 1A group (Fig. 3). In the present study, no Africa 1C related virus (SM2) was detected. According to the discrete and continuous geographical models developed (Figs. 3, 4, SM1, 2 and 3), the rabies genetic clustering correlates to biogeography (Fig. 3a) and the RABV detected in Uganda are related to several phylogenetic lineages coming from several African biogeographical zones (Figs. 3 and 4). Isolates identified as 13RS266_29/Jackal/2011/Kabongo/Uganda, 13RS266_3/Dog/2012/Kabarole/Uganda, 13RS266_23/Bovine/2012/Moyo/Uganda clustered with Tanzanian carnivore isolates (DQ900555/Dog/2001/Tanzania, DQ900567/spotted_hyena/2004/Tanzania, DQ900554/Dog/1998/Tanzania) which belongs to Africa 1B lineage circulating in East and Southern Africa.

A dog isolate from Moyo district (13RS266_9/Dog/2010/Moyo/Uganda), strongly clustered with a Nigerian strain (U22488_Nigeria_RVU22488) (Fig. 3 and SM1). Together they form a clade that clearly diverges from the root of all other Africa 1B.

Six Ugandan virus isolates from Northern and South Western Uganda (13RS266_10/Dog/2012/Moyo/Uganda, 13RS266_14/Dog/2010/Moyo/Uganda, 13RS266_20/Dog/Ntoroko/Uganda, 13RS266_1/Goat/2010/Ntoroko/Uganda, 13RS266_30/Bovine/2012/Kabarole/Uganda, 13RS266_8/Dog/2010/Moyo/Uganda) were close to Sudanese isolates from dogs (AY502129/Dog/2001/Sudan, AY502131/Dog/2002/Sudan, AY502126/Dog/2001/Sudan) belonging to the lineage Africa 1A circulating in dogs from Northern and eastern Africa (Fig. 3 and SM1).

Virus strains of the African 1B group detected in Uganda (mostly from Moyo, Ntoroko and Kabarole districts, 2010 to 2012), clustered as a sister-clade of RABV that circulated in Sudan since 2001 and appear to share a common ancestor among the Afrotropical-Ethiopia-Somalian Biogeographical zone (i.e. Ethiopian clade) (Fig. 3). The relaxed molecular clock analysis suggests that the diversification of this clade occurred during the second half of the twentieth century. Phylogenetic analyses link the circulation of one lineage 1B in Uganda (detected from 2010 to 2012 in all localities sampled) to Tanzanian viruses and the episodic presence of a second lineage (in 2010 in Moyo), also to a Tanzanian lineage.

The study had limited funding and time for the researcher to cover a wider scope of these two districts. The study design was retrospective and descriptive. The study implementation was focused on understanding the risk and identifying Rabies control and prevention gaps for the population. Also, the study sites where identified to assess the transborder risk of rabies circulation.