Canine rabies is endemic among domestic dogs in the South African provinces of KwaZulu Natal (KZN) and the Eastern Cape (EC) from where about 6–29 human rabies cases (mean 14 cases), most of whom are young children are reported annually [Reference Bishop1]. Canine rabies is a relatively new disease in the Republic of South Africa (RSA) and two major epidemics have occurred among the local dog populations of the eastern coastal regions of the country during the last 50 years. The first of these epidemics entered and manifested in KZN during 1964. The epidemic was thought not to have reached the EC, and was eradicated by 1968 through mass vaccination and strict dog control [Reference Bishop1, Reference Swanepoel, Coetzer and Tustin2]. New evidence suggests that this epidemic did in fact reach the EC, where it has persisted ever since this period [Reference Coetzee and Nel3]. Following the elimination of rabies from KZN by 1968, a new epidemic emerged in the northern parts of this province in 1976. This epidemic coincided with the outbreak of civil war in bordering Mozambique and the fleeing of refugees across the international border with RSA (northern KZN). This second epidemic has proven to be intractable, despite the vigorous control measures that have been implemented by the South African Directorate of Veterinary Services [Reference Bishop1, Reference Swanepoel, Coetzer and Tustin2].

We have previously studied the molecular epidemiology of this radiating epidemic, concentrating on 142 rabies virus (RABV) specimens that were obtained from domestic dogs and livestock from the KZN and EC provinces between the calendar years 2003–2004 [Reference Coetzee and Nel3]. The study targeted a 592 nucleotide region that encompassed the carboxyl terminal domain of the glycoprotein (G, cytoplasmic domain) and the G–L intergenic region (L representing the downstream ‘large’ viral polymerase gene) that constitutes a highly variable portion of the RABV genome. This region is suited for distinguishing closely related RABV variants in localized geographic domains [Reference Coetzee and Nel3–Reference Nel5] and was successfully applied towards the identification of regional RABV variants that were cycling during the study period [Reference Coetzee and Nel3]. The resulting phylogenetic analyses demonstrated that RABV in the KZN and EC provinces could be divided into subfamilies, clusters and groups that were unique to these provinces and to specific localized geographical areas within these provinces. This study represented the most complete molecular description of the canid rabies epidemic in the affected provinces to date and the generated sequence database is likely to be useful in future surveillance and epidemiological investigations [Reference Coetzee and Nel3]. Such epidemiological data can, for example, be applied to monitor the expansion of the epidemic into unaffected regions along the eastern coast of the country and in the same way the effectiveness of control strategies may be monitored. We have used the database to retrace human case histories where a case history was either incomplete or absent. Five human RABV specimens recovered from children aged <14 years from the KZN and EC provinces during 2002–2003 were studied and the main regional and local variants responsible for each infection were identified. The resulting evidence in one of these cases also suggested a passive non-bite transmission of RABV from an infected dog to a young child.

During 2002–2003, 21 human cases of rabies were laboratory-confirmed in South Africa [6, 7]. Of these, four specimens from KZN (KZNhmSPU03.15, KZNhmSPU03.77, KZNhmSPU03.272, KZNhmSPU02.326) and one from the EC (EChmSPU03.48) were selected for further study (Table). Inclusion of these specimens was based on the availability of post-mortem brain specimens and partially completed case histories as documented by field epidemiologists and state pathologists on specimen submission forms that are available from the Special Pathogens Unit at the National Institute for Communicable Diseases (NICD, RSA). Brain specimens were sampled at autopsy and were stored at −20°C in 50% glycerol-phosphate-buffered saline (PBS) without further passage. RNA extraction, RT–PCR, nucleotide sequencing and phylogenetic analysis was performed as previously described [Reference Coetzee and Nel3]. Phylogenetic analysis indicated that KZNhmSPU03.15, KZNhmSPU03.77, KZNhmSPU03.272, KZNhmSPU02.326 and EChmSPU03.48, clustered into viral clusters KZN/A/V1, KZN/A/V5, KZN/A/V3/Cl1, KZN/A/V6 and EC/A/V2, respectively (Fig. 1). Clustering of RABV into these groups was in most cases supported by bootstrap values above 70%. The approximate intrinsic sequence identity viruses belonging to each of the clusters varied from 99·3% to 99·9%, whereas the mean sequence divergence between these clusters ranged from 1·0% to 2·2%. The average sequence identity for the human RABV specimens to the clusters into which they grouped ranged from 99·94% to 99·99% (p distances calculated in MEGA 3.1 [Reference Kumar, Tamura and Nei8], data not shown). All of the identified sequence clusters correlated with the general geographical regions where the cases were reported from, namely Lower Umfolozi, Port Shepstone, Tugela Ferry, Port Shepstone, and regions surrounding Cofimvaba – in that order (Figs 2 and 3). It was also of interest to note that within each of the sequence clusters from KZN, one or more animal viruses (KZN/A/V1=20; KZN/A/V3/CL1=2; KZN/A/V5=1; KZN/A/V6=7) that shared 100% G–L sequence identity with the respective human viruses, could be identified.

Table. Case histories of selected human infections which occurred in the KwaZulu Natal (KZN) and Eastern Cape (EC) provinces during the calendar years 2002–2003

* Laboratory reference numbers: human viruses which were obtained from the National Institute for Communicable Diseases (NICD, RSA) are named using an SPU (Special Pathogens Unit) designation.

† The approximate regions from where specimens were obtained, have been indicated by using a grid reference system, as implemented on the Allerton Regional Veterinary Laboratory (KZN) specimen submission forms.

Fig. 1. Neighborhood-joining tree of 82 nucleotide sequences of the cytoplasmic domain of the glycoprotein and G–L intergenic region, for selected canine, domestic livestock and human rabies viruses from KwaZulu Natal (n=71) and the Eastern Cape (n=11) provinces of South Africa [Reference Coetzee and Nel3]. Horizontal branch lengths are proportional to the similarity of the sequences within and between groups, with the scale indicating the amount of nucleotide sequence divergence in substitutions per site. The vertical lines are provided for purposes of clarity only. The cytoplasmic domain and G–L intergenic sequence region from a bat-eared fox specimens (o491.98) was used as reference sequence to root the tree, with virus numbers being preceded by a prefix indicating the geographic region (KZN, KwaZulu Natal; EC, Eastern Cape) as well as host species of isolation (dg, dog; cw, bovine; gt, caprine; hm, human). Symbols are used to denote subfamily and group divisions on the phylogenetic tree, as well as the geographical distribution of these variants on maps of the KZN and EC provinces in Figures 2 and 3. Human specimens are boxed in the phylogenetic tree.

Fig. 2. A map of the KwaZulu Natal (KZN) province demonstrating the approximate geographic origin of canid rabies virus variants [Reference Coetzee and Nel3]. Symbols correspond to those used for the respective viral groupings on the phylogenetic tree in Figure 1. The approximate geographic regions where human cases originated from are indicated by arrows.

Fig. 3. A map of the Eastern Cape (EC) province demonstrating the approximate geographic origins of canid RABV variants [Reference Coetzee and Nel3]. Symbols correspond to those used for the respective viral groupings on the phylogenetic tree in Figure 1. The approximate geographic region where the human case originated from is indicated by an arrow.

Rabies persists as a significant human and veterinary public health problem in RSA and much of the developing world. Some of the reasons for this include competing public health priorities, limited resources for veterinary control measures, shortages and prohibitive costs of post-exposure biologics, poorly informed communities, inadequately trained health and veterinary staff and inaccessible health-care facilities [Reference Bishop1]. Most victims of the disease only present to primary health-care providers when already in the terminal stages of the disease. Further, the majority of victims are children who because of their height, inquisitive nature, interest in animals and inability to protect themselves, are particularly vulnerable to bite exposures from domestic dogs, especially to the face and head. Such injuries are often associated with shortened disease incubation periods [Reference Bishop1]. Exposure of young children can complicate the task of reconstructing case histories, which is an essential step towards identifying possible contact routes and for determining where appropriate measures such as information campaigns and vaccination of domestic dogs should be undertaken. Often, young children are unable to avoid potential risk situations, are unaware of potential exposures and are unable to effectively communicate such potential exposures. Phylogenetic analysis of generated nucleotide sequence data can prove useful for retracing human case histories in regions where variant distributions have been delineated and has in the past been applied to trace the importation of human rabies cases into non-endemic countries [Reference Fooks9, Reference Smith10], to trace the origin of cases in which there was no history of a bite exposure [Reference Messenger, Smith and Rupprecht11], and to identify foreign imported rabies cases that demonstrated extended incubation periods [Reference Fooks9]. Without exception, the RABV studied here did conform to the phylogenetic grouping specific to viruses found to be circulating in the geographical domain associated with each case. Indeed it was found that human viruses in each of the respective KZN sequence clusters were identical to one or more animal viruses that were obtained from proximal geographic locations. This finding further highlights the fact that human exposures in the province occur in the context of highly localized outbreaks of specific RABV variants among domestic dogs. If this had not been the case, aspects of translocation or introduction of newly recognized variant(s) would have warranted further investigation.

Of particular interest is the case of a 6-year-old boy, case KZNhmSPU02.326. For this child there had been no history or evidence of a bite exposure or other close contact with any dog prior to him developing rabies. In retrospect, however, the next-door neighbour's dog had in fact died of unknown causes during the same relative time period, although rabies was not suspected nor considered in the animal. This unfortunate and injudicious oversight also meant that brain specimens were not available for our retrospective analysis. Nevertheless, based on phylogenic identity to virus variants which circulated in the region where this exposure occurred, it is highly likely that a scratch or lick, or other exposure to saliva from this dog was responsible for the transmission of variant KZN/A/V6 to the child. Effective public health practices require that all rabies cases should be followed up and other potential exposures identified and managed. In cases such as these, where exposures go unnoticed, the use of molecular genetic tracking and locating of the source should be of particular importance if provided in a timely manner.

The KZN and EC provinces are endemic for the rabies-related viruses Lagos Bat virus (LBV) and Mokola virus (MOKV), while the mongoose biotype of classical rabies virus circulates among herpestids elsewhere in RSA [Reference Swanepoel, Coetzer and Tustin2, Reference Nel5, Reference Nel12]. Together with the recent isolation of LBV from a marsh mongoose in KZN, the possibility of potentially productive infectious cycles among terrestrial animals in these regions, which could lead to human exposures, has been raised [Reference Markotter13]. From this perspective, our database as well as the methods that are described here should be continually expanded as new cases occur and should include not only canine rabies viruses, but indeed all the lyssaviruses of the region. Finally, rabies is also escalating in other parts of South Africa and the larger subcontinent – indeed a record number of human rabies cases were identified in South Africa in 2006 due to an outbreak among dogs in the Limpopo Province [Reference Paweska, Leman and Blumberg14]. It is clear that rabies – a completely preventable disease – not only needs to be re-prioritized, but that improved tools can and should be applied within rabies control programmes.