Every year an estimated 59,000 people die from rabies (Hampson et al., 2015), a viral infection transmitted primarily by domestic dogs in low- and middle-income countries (LMICs). While human rabies encephalitis remains incurable, the disease is readily preventable if post-exposure prophylaxis (PEP) is promptly administered to bite victims upon exposure (World Health Organization, 2018b). Moreover, mass dog vaccination has eliminated dog-mediated rabies from high-income countries and much of the Americas (Vigilato et al., 2013). Yet, in most African and Asian countries there has been little investment in dog vaccination and rabies circulates unabated (Hampson et al., 2015). A global goal to eliminate human deaths from dog-mediated rabies by 2030 (‘Zero by 30’) is now advocated (Minghui et al., 2018), with recommendations to scale up mass dog vaccination.

Although dog vaccination can eliminate dog-mediated rabies, there are challenges to achieving this goal. In most rabies endemic countries in sub-Saharan Africa, dog vaccination campaigns have been sparse and localised (World Health Organization, 2018a). Moreover, the high reproductive rates and short lifespan of dogs in many LMICs quickly lead to drops in vaccination coverage, with repeat campaigns required to maintain coverage (Davlin and Vonville, 2012). The virus can easily spread in dog populations that have low and heterogeneous vaccination coverage (Mancy et al., 2022) and incursions leading to outbreaks are commonly reported (Bourhy et al., 2016; Zinsstag et al., 2017; Rysava et al., 2020), often facilitated by human-mediated movement of dogs incubating infection (Townsend et al., 2013b; Tohma et al., 2016). This situation is compounded by weak surveillance which hinders effective outbreak response and poses a challenge for monitoring progress towards elimination, including how to determine disease freedom (Nel, 2013).

Across the African continent there are very few documented examples of elimination of dog-mediated rabies. We found just four papers reporting locations on the continent with potential interruption of transmission by dog vaccination; the cities of N’Djamena, Chad (Zinsstag et al., 2017) and Harare, Zimbabwe (Coetzer et al., 2019), Serengeti district in northwest Tanzania (Cleaveland et al., 2003) and KwaZulu-Natal province in South Africa (Sabeta and Ngoepe, 2018). In all four locations, endemic circulation has since re-established, with resurgences explained by movement of infected dogs from surrounding areas after dog vaccination campaigns lapsed. The importance of reintroductions in maintaining rabies circulation is further highlighted from long-term surveillance from Bangui, the capital of the Central African Republic (Bourhy et al., 2016) and from long-term contact tracing in Serengeti district, Tanzania (Mancy et al., 2022). Genomic surveillance can potentially play a role in differentiating rabies introductions from undetected sustained transmission, and thus in confirming or refuting rabies elimination and therefore targeting of control efforts. However, sequencing of rabies viruses also remains limited in Africa.

Dog-mediated rabies is endemic in East Africa where thousands of human rabies deaths occur each year (Hampson et al., 2019). Rabies has circulated on Pemba Island, off mainland Tanzania, since the late 1990s. Dog vaccinations on Pemba first began in 2010, with a small-scale campaign conducted by the animal welfare organisation, World Animal Protection (formerly WSPA). Over the next 5 years, a rabies elimination demonstration project, funded by the Bill & Melinda Gates Foundation, coordinated by the World Health Organisation and led by the Tanzanian government, was implemented across southeast Tanzania, including Pemba (Mpolya et al., 2017). Here, we show how these efforts led to rabies elimination, while highlighting how introductions pose challenges to achieving and maintaining rabies-freedom even on a small, relatively isolated, island. Our study is the first to confirm rabies elimination from an African setting, including in response to reintroduction, through quantifying case detection. Using rigorous contact tracing, we identified chains of transmission within the domestic dog reservoir informed by in-country whole-genome sequencing (the first example in Africa) and cross-species transmission from domestic dogs to humans. This enabled us to estimate the public health burden and associated cost-effectiveness of both post-exposure vaccination and dog vaccination, as well as their combined use, in achieving and maintaining rabies freedom on Pemba. Our findings illustrate the critical need to holistically link surveillance with public health and veterinary interventions to cost-effectively reduce the burden of zoonotic pathogens. This case study provides timely lessons given the global strategic plan to eliminate dog-mediated human rabies by 2030.

Rabies was endemic on Pemba in 2010 at the start of the study. That year we traced 32 human rabies exposures, 33 rabid dogs and three human rabies deaths diagnosed from clinical signs and history of exposure (6.77 exposures and 0.63 deaths/ 100,000 persons and 10.5 cases/ 1000 dogs). Initial dog vaccination implemented as part of a rabies elimination demonstration project in 2011 achieved only low and heterogeneous coverage (13% in 2011, ranging from 7% to 20% across districts), but by 2014 campaigns were island-wide and achieved better coverage (mean 50%, range 46–60%, Figure 1). Correspondingly, human rabies exposures and dog rabies cases declined each year to just 2 each in 2014. The effective reproduction number, R_e, also declined from around 1.5 in 2010 to <1 in 2014 (Figure 2). No human rabies exposures, deaths or animal cases were detected from May 2014 until July 2016.

Figure 1

Download asset Open asset

Figure 2 with 1 supplement see all

Download asset Open asset

In August 2016, an influx of bite patients was seen in hospitals on Pemba. By the year end, we had traced 35 human rabies exposures and 27 dog rabies cases. In response to this outbreak, the Ministry of Livestock and Fisheries Development initiated dog vaccination in shehias with recorded dog cases, but because the outbreak spread rapidly, islandwide dog vaccination was reinstated. In 2017, we traced three human rabies deaths, 126 rabies exposures and 62 rabid dogs (26.6 exposures and 0.63 deaths/100,000 people, and 19.6 cases/1000 dogs). High vaccination coverage was achieved consecutively over subsequent annual dog vaccination campaigns from 2017 to 2020 (median 61%, range 46–78% in 2019, Figure 1). Incidence rapidly declined from the 2017 peak with 19 human rabies exposures and 8 dog rabies cases detected in 2018. At the start of the outbreak R_e was high (>1.5), but subsequently declined to <1, with all transmission interrupted by October 2018 (Figure 2). No human rabies exposures, deaths or rabid dogs have been identified since (as of September 2022).

Phylogenetic analyses indicated considerable viral diversity from 2010 to 2014 (Figure 3). Pruning by time split the reconstructed tree into two transmission chains, from 2010 to 2014 and 2016 to 2018, respectively. Integrating the phylogenetic data further split the transmission chains into varying sizes associated with each sampled lineage, consequently resolving five distinct transmission chains from 2010 to 2014 (Figure 4, Video 1). During this period we detected approximately 54% of dog rabies cases circulating on the island (95% credible intervals (95% CI) 46.4–62.0%, 92 of an estimated 171 rabid dogs 95% CI: 148–198 rabid dogs, Figure 5). Further pruning by the dispersal kernel distance threshold created an additional 3–4 unsampled transmission chains and 5–6 orphaned cases, likely because links connecting these cases were either missed or human-mediated. At least one divergent lineage (reported as Tz5 in Brunker et al., 2020) is known to have circulated during this period. Figure 4—figure supplements 3–5 show in grey cases that were not grouped by the current phylogenetic assignment.

Figure 3 with 2 supplements see all

Download asset Open asset

Figure 4 with 6 supplements see all

Download asset Open asset

Figure 5 with 1 supplement see all

Download asset Open asset

Video 1

Download asset

Using RABV-GLUE to obtain an alignment of all sequences from the Cosmopolitan-AF1b minor clade up to the year 2017 (inclusive) regardless of genome position or length resulted in a dataset of 2557 sequences (partial and whole-genome) from 21 countries (all sub-saharan Africa, aside from one Thailand sequence) spanning the years 1980–2017 (Figure 3—figure supplement 1). This provided a much wider geographic and temporal context than only the WGS sequences enabling the placement of 2016/17 Pemba outbreak sequences into the context of known background diversity. N and G genes were the most commonly sequenced, with 1042 and 1876 sequences respectively (including WGS), and both gene datasets were used to contextualise Pemba sequences. There was wide variation in the portion of each gene covered and the length of sequences (N: from 203 to full length 1353 basepairs (bp); G: from 277 to full length 1575 bp). The N gene dataset constituted a wider and more even geographic distribution (18 countries), whereas the G gene (16 countries) data was predominantly from South Africa and Tanzania.

Bootstrap values were generally low across the phylogenies, most likely due to the use of short, variable length sequences and therefore should be interpreted with caution. We ‘zoomed in’ on clusters within these subtrees to identify the closest relatives to Pemba outbreak cases (Figure 3—figure supplement 2). For Pemba cluster 1 (Figure 3B, and Figure 3—figure supplement 2a, b), sequences were most closely related to sequences from the Serengeti District in northern Tanzania according to both N and G gene datasets. Whereas Pemba cluster 2 sequences (Figure 3C, Figure 3—figure supplement 2c, d) share a common ancestor with N gene sequences from Zanzibar, a neighbouring island, from the same period (2016/17). This suggests a possible link between rabies outbreaks on these islands and/or a common source of introduction.

Viruses sequenced from the outbreak starting in 2016 belonged to two distinct phylogenetic lineages (Figure 3). The time-scaled phylogeny pointed to two independent introductions taking hold and spreading widely, i.e., not continued transmission of viruses circulating previously. Although our estimates of case detection were higher during this outbreak, at 69% (95%CI: 59.4–81.6%, Figure 5B), i.e. 97 out of an estimated 140 rabid dogs (95%CI: 119–163 cases in outbreak), a few connections linking unsampled transmission chains were still likely missed, with pruning by the dispersal kernel threshold identifying 2–5 unsampled lineages and orphaned cases (Figure 4—figure supplements 2–5). Our estimates of R_e over time were consistent irrespective of the transmission tree summary, pruning algorithm or whether reconstructed with or without phylogenetic information (estimates not shown since indistinguishable). Individual variation in rabid dog biting behaviour was identified from contact tracing and inferred for case estimates of R_e (Figure 2 also shows variation in persons exposed by individual rabid dogs). On average secondary cases per case declined with higher vaccination coverage in the locality when each case occurred (Figure 2—figure supplement 1). While there was uncertainty in exact progenitor assignments, transmission chain assignments were consistent, and most cases had only a few highly plausible progenitors (>10% chance of being assigned progenitor, Figure 4—figure supplements 3–6).

Our approach to case detection inferred from the transmission tree reconstructions were also generally robust, irrespective of the tree summary, pruning algorithm and for the amount of data available. However, our approach tended to underestimate detection probabilities (usually within 10%, Figure 5A) at the levels inferred for Pemba (50–70%). That withstanding, we show that given the estimated case detection probabilities, we likely detected most transmission chains comprising more than two cases (Figure 5C) and missed only dead-ends or very short chains of transmission.

Of the bite patients presenting to the island’s four hospitals from 2010 to 2014 (n=117), a large proportion were bitten by probable rabid dogs (45–72% depending upon the status of unclassified biting dogs), while only a few patients that were bitten by apparently healthy dogs sought care during this period (6.6–12.8 per year, or 1.4–2.7/100,0000 /year). Based on the probability of rabies progression (Hampson et al., 2019) and the occurrence of three human rabies deaths in 2010 (Table 1), we estimated that 10–31 rabid bite victims did not receive complete or timely PEP and through contact tracing we identified 21 such rabies exposures. The total rabies exposures that we detected from 2010 to 2014 (63-94) were within expectations from triangulating case detection and rabid dog behaviour (Hampson et al., 2016; 65 exposures, range 46–86), and consistent with a 0.66–0.89 probability of rabies-exposed bite victims receiving adequate PEP. During the 2016–2018 outbreak we traced 39 rabid bite victims who did not obtain adequate PEP (late and/or incomplete) and estimated that exposures received appropriate PEP with probability 0.72–0.78 (with the three deaths that occurred early in the outbreak suggesting around 10–31 rabid bite victims did not receive adequate PEP). Probable exposures per rabid dog were higher during the 2016–2018 outbreak than from 2010 to 2014 (1.3 vs 0.34-0.51, both adjusted for case detection) driven in part by variability in dog biting behaviour; two rabid dogs in 2017 each bit more than 10 people (Figure 2).

Reasons reported for lack of, or inadequate PEP varied (detailed for deaths in Table 1). No PEP shortages were reported whilst PEP was provided for free during the elimination demonstration project (2011–2014). But, at the start of the outbreak in late 2016 patients had to buy PEP (~$12.9 per vial or >$38 for a complete course) and one child bitten in early 2017 by a confirmed rabid dog did not receive PEP due to a shortage. Although too late to be effective, health authorities sought PEP from Zanzibar when the child presented with symptoms, but none was available. In desperation the family took the child to the mainland but with no rabies treatment options they were advised to return home where the child died on arrival. Following the child’s death, Zanzibar’s Ministry of Health imported PEP and reinstated free-of-charge PEP provisioning. This policy change and sensitization around the outbreak likely contributed to increased health seeking and understanding of the critical need for timely PEP. Contact tracing revealed that 17.5% of rabies exposures were not aware of the importance of PEP early on (2010–2014) compared to <4% during the outbreak (2016–2018) and similarly around 20% of rabies exposures early on (2010–2014), reported not being advised by health workers to obtain PEP, declining to 3% during the outbreak (2016–2018). Patients presenting for healthy dog bites also increased during the outbreak to ~37 /year (7.8/100,000 vs 1.4-2.7/100,000 previously). In the decision tree we used estimates of the probabilities of rabies-exposed bite victims starting and completing PEP for the period 2010–2014 when most patients paid for PEP, and 2016–2020 when most patients received free PEP. Probabilities for starting and completing PEP were 0.667 and 0.397 respectively for 2010–2014 and 0.783 and 0.84 for 2016–2020 respectively (Figure 6).

Figure 6

Download asset Open asset

Figure 7

Download asset Open asset

The cost of a complete intramuscular post-exposure vaccination course (4-dose Essen regimen) was approximately $56 versus $25 for an intradermal course (updated Thai Red Cross). Over the 11 years of the study around $17,800 was spent on PEP for 542 bite patients who received a combination of intramuscular and intradermal regimens and had varying levels of compliance. We estimated that this PEP prevented around 42 rabies deaths (95% confidence intervals: 32–55) costing around $424 per death averted. From 2019 onwards, in the aftermath of the 2016–2018 outbreak when all transmission had been interrupted, approximately $876 was spent annually on PEP for patients presenting with bites from healthy dogs (Figure 1), i.e., precautionary expenditure post-elimination. Island-wide dog vaccination cost approximately $12,122 per campaign ($13,145 for the campaign that reached most dogs), with a cost of $6.5 per dog vaccinated (range: $4.2–10.8 depending on the campaign). Dog vaccination campaigns interrupted transmission in the dog population within four years of implementation, first in 2014 and again in 2018. However, the lapse in dog vaccination from 2014 allowed the two introductions in 2016 to spread widely.

We parameterized a probabilistic decision tree model and projected rabies incidence, exposures, and deaths under counterfactual scenarios (Figures 6 and 7). We estimated that without dog vaccination and with PEP charged to patients (i.e. the status quo prior to the rabies elimination demonstration project) around 27 deaths (95% prediction intervals (95% PIs): 16–39) would occur on Pemba over a 10-year time horizon. On average 48 deaths (95% PIs: 31–67) would be prevented by PEP, at a cost of $300 per death averted (95% PIs: $263–374, with costs discounted at 3%) incremental to a counterfactual without PEP, i.e., in the absence of interventions, 75 human rabies deaths would be expected to occur over ten years on Pemba. Providing PEP for free to patients (as during the rabies elimination demonstration project and by Pemba’s government from 2017 onwards) was projected to prevent an additional 10 deaths at a cost of $256 per death averted (95% PIs: $217–333), but still result in 17 rabies deaths (95% PIs: 9–26) over the ten years, with intradermal regimens always more cost-effective than intramuscular regimens. Introducing and sustaining mass dog vaccination, whilst charging for PEP, was projected to prevent 20 deaths relative to the status quo (68 deaths averted overall, 95% PIs: 45–92) costing $1,684 per death averted (95% PIs: $1,264–2,515). Dog vaccination together with free PEP was projected to result in fewest deaths (4 overall, 95% PIs: 1–9), with no deaths after year four (Figure 7), and preventing 71 deaths overall (95% PIs: 46–97) at a cost of $1,657 per death averted (95% PIs: $1,228–2,526). Since dog vaccination interrupts transmission, we project that routine dog vaccination would mitigate ongoing risks from introductions and keep Pemba rabies-free, and thus prevent over 300 rabies exposures over the ten years (95% PIs: 263–401) sparing around 30 families each year from rabid dog bites and the anxiety of needing to urgently obtain life-saving PEP. These results remained robust with and without discounting costs (results not shown).

Dog-mediated rabies is the quintessential zoonotic disease requiring coordinated public health and veterinary interventions as part of a One Health approach to end unnecessary suffering and deaths. Inequities in access to both human and animal vaccines manifest in the continued high burden of rabies in neglected communities around the world. Our study quantifies dog-mediated rabies transmission in an African setting, illustrating how rabies incidence in domestic dogs translates to human rabies exposures and how limitations in provisioning post-exposure vaccines results in human deaths. On Pemba, endemic rabies led to many exposures and three deaths in 2010, the year that our study began. Over the following four years consecutive islandwide dog vaccination campaigns were undertaken during a rabies elimination demonstration project. Initially, only low vaccination coverage was achieved (Figure 1), but prevention efforts improved after implementation challenges were overcome, including lack of dog vaccination experience and poor health seeking, conflated by expensive and inaccessible PEP. By 2014, transmission was interrupted. Unfortunately, two independent introductions to Pemba in 2016, at a time when vaccination coverage in the dog population was low, seeded a large outbreak causing three further rabies deaths in 2017. Attempts to respond locally were ineffective, until Pemba’s government re-established dog vaccination islandwide, after which rabies was rapidly eliminated. The island has remained rabies-free since October 2018.

Accumulating evidence illustrates how metapopulation dynamics maintain dog-mediated rabies via endemically co-circulating viral lineages (Mancy et al., 2022; Bourhy et al., 2016). Despite Pemba being a relatively isolated island with a small dog population, genomic data revealed considerable RABV diversity, likely arising from historical introductions, (Brunker et al., 2015). The rapid outbreak spread from two contemporary introductions in 2016 highlight the fragility of elimination. While dog rabies remains uncontrolled in nearby populations, reintroduction risks are high (Bourhy et al., 2016; Zinsstag et al., 2017; Townsend et al., 2013b; Tohma et al., 2016; Sharma et al., 2010). Re-emergence is most likely if dog vaccination coverage is low, causing major public health and economic consequences (Zinsstag et al., 2017; Tohma et al., 2016; Windiyaningsih et al., 2004; Castillo-Neyra et al., 2017). Introductions may be reduced through improved border control, but informal human-mediated movement of dogs is not easy to regulate. Scaling up coordinated dog vaccination should suppress the source of introductions and accelerate elimination, accruing and sustaining long-term benefits across much larger populations. While our study from this small island dog population represents a best-case scenario, examples from Latin America show dramatic contractions of dog-mediated rabies when dog vaccination is scaled up and sustained. The last dog-mediated rabies foci on the continent remain only in very poor communities where dog vaccination has been inadequate (Vigilato et al., 2013; Rysava et al., 2020).

Our detailed contact tracing data from Pemba contrasts with very weak routine rabies surveillance in both humans and animals throughout much of Africa (Nel, 2013). Low case detection leads to underestimation of disease burden, lack of prioritisation, and difficulty ascertaining impacts of control, including whether disease has been eliminated, or is circulating undetected. The high case detection on Pemba generated confidence that elimination was achieved, twice (Townsend et al., 2013a). The subsequent re-emergence emphasises the need to maintain both surveillance and vaccination coverage where the risk of introductions from connected populations remains. The whole-genome sequences generated in this study further revealed the underlying metapopulation dynamics of rabies circulation. By identifying introductions to Pemba and resolving their role in further spread it was possible to rule out sustained undetected transmission as the cause of re-emergence.

Although contact tracing was intensive compared to routine rabies surveillance which is passive and ad hoc in most African countries, much more intensive approaches are used in high-income countries in response to rabies incursions (Collective french multidisciplinary investigation team, 2008), highlighting how dog-mediated rabies is considered a public health emergency. Only one person (KL) completed all contact tracing on Pemba, while also covering other mainland sites (Lushasi et al., 2021), in coordination with local livestock field officers, illustrating the capacity of a dedicated epidemiologist/ One Health specialist focusing on rabies. Contact tracing was tractable on Pemba given the relatively low incidence of bite patients (in contrast to some Asian settings, Hampson et al., 2019), of which a high proportion were rabies exposures identifiable upon presentation to health facilities. As well as strengthening One Health capacity and intersectoral relationships, contact tracing is a vital tool for investigating emerging infectious diseases, such as Ebola (Faye et al., 2015). Contact tracing could be a useful component of rabies control programmes when approaching elimination as caseloads become manageable. Without enhanced surveillance during the endgame, incursions would be more likely to go undetected until human deaths occur (Townsend et al., 2013b). Other less intensive approaches, such as integrated bite case management, may be more feasible if used to first increase case detection, strengthen intersectoral working and build technical competence in One Health working generally (Swedberg et al., 2022). Genomic approaches are also increasingly affordable, and capacity for genomic surveillance is growing, accelerated over the course of the SARS-CoV-2 pandemic. For rabies, genomic approaches have potential to enhance the information that can be gleaned from routine surveillance and inform elimination programmes, which are likely to experience such introductions as they progress.

There were limitations in our reconstructions of transmission events despite high case detection. As evident from the tree summaries (Figure 4—figure supplements 2–6), we were unable to identify exactly who infected whom within transmission chains, given the low mutation rate of rabies virus and the short intervals between cases. However, we have confidence in the general transmission tree topology and in plausible progenitors within transmission chains (most cases had only a few candidates). Moreover, relevant epidemiological inference, such as estimates of R_e and of outbreak sizes were robust even with uncertainties in transmission pathways and under alternative pruning algorithms. Thus, while improvements in methods to reconstruct transmission trees from both epidemiological and genetic data are warranted, particularly for further characterising spatiotemporal variation in case detection (we assumed all cases had equal probability of being detected), our methods proved valuable. We caveat that while we retrospectively identified a high proportion of cases from their cases histories, we were less successful in recovering samples (~10% of identified cases overall, increasing to >12% during the outbreak). Nonetheless, the genomic data provided key insights that would not have been evident with only epidemiological data: first that rabies was not circulating undetected from 2014–2016 (rabies only re-emerged because of external introductions); second that two introductions occurred in 2016 rather than the outbreak arising from a single introduction (highlighting the non-negligible risk from circulation elsewhere); and finally that it is highly unlikely that other introductions after 2016 established onward transmission (whereas the diversity of lineages circulating pre-2014 indicated multiple introductions previously, Brunker et al., 2015). Improved sample recovery and sequencing will be valuable for refining methods, but the relatively low mutation rate of rabies may ultimately limit fine-scale inference, on for example, who infected whom.

Our analyses highlight the cost-effectiveness of PEP as an emergency medicine critical for rabies prevention. We estimated a very low cost per death averted for free PEP provisioning on Pemba (Figure 7), even when considered incrementally to the status quo where PEP is charged to patients. Our estimate from Pemba is amongst the highest cost-effectiveness estimates of PEP from across Gavi-eligible countries (Hampson et al., 2019) (translating to a cost of just $13 per DALY averted) and results from the high proportion of bite patients presenting with rabies exposures rather than bites from healthy dogs. In settings with more patients seeking care for healthy dog bites, PEP cost-effectiveness declines, although this can be slightly offset by increased vial sharing opportunities under intradermal dose-sparing regimens. Even though PEP is an essential emergency medicine, PEP does not address the suffering caused from injuries inflicted by rabid animals and is insufficient to protect the entire at-risk population. Our study shows how, in practice, lack of awareness, expense and supply issues still prevent access to these emergency vaccines for marginalised populations.

In contrast to PEP sustained mass dog vaccination reduces the risk of exposure and by interrupting transmission in the reservoir can achieve the equitable goal of elimination. Mass dog vaccination inevitably comes at a higher cost per death averted (Figure 6) particularly given the relatively high cost per dog vaccinated in this setting. Nonetheless compared to other health interventions (Bertram et al., 2021), this One Health approach remains extremely cost-effective. In denser, more connected populations than Pemba rabies elimination is likely to take longer and be more fragile, while conversely the cost per dog vaccinated is likely to reduce in areas with larger dog populations and with opportunities for optimising the delivery of dog vaccination. Our cost-effectiveness estimates lie within expectations for countries in sub-Saharan Africa (Hampson et al., 2019), but these considerations limit their transferability. To improve health economic models, the relationship between dog vaccination coverage and risk reduction needs to be better quantified, and research is needed on health seeking behaviours following bites by both healthy and rabid dogs that impact cost-effectiveness. Moreover, realised cost-effectiveness depends on the stochastic nature of outbreaks and the degree to which interventions are delivered as intended, including how dog vaccination coverage is maintained, since vaccination campaigns can often lapse, as seen from Pemba. The COVID-19 pandemic highlights how such disruption can severely set back rabies programmes (Kunkel et al., 2021; Raynor et al., 2021, Nadal et al., 2022).

We conclude that the investment needed for a One Health approach, to support access to life-saving emergency vaccines for bite victims and to achieve and maintain rabies freedom in source populations through dog vaccination is very cost-effective and can bring rapid success. Lessons from Pemba should build confidence in the feasibility of eliminating rabies elsewhere on the African continent but highlight the importance of sustaining commitment. Coordinated dog vaccination over sufficiently large scales will have the greatest and most long-lasting impacts in equitably tackling this preventable disease.

Code to reproduce the analyses together with deidentified data are available from our public Github repository https://github.com/boydorr/PembaRabies and archived on Zenodo DOI: https://doi.org/10.5281/zenodo.7922464. Sequences are deposited on Genbank.

1. 1. Lushasi, et al

   (2023) Zenodo

   boydorr/PembaRabies: Integrating contact tracing and whole-genome sequencing to track the elimination of dog-mediated rabies: an observational and genomic study.

   https://doi.org/10.5281/zenodo.7922464

1. 1. Bertram MY
   2. Lauer JA
   3. Stenberg K
   4. Ralaidovy AH
   5. Edejer TTT

   (2021) Progressive realisation of universal health coverage in Low- and Middle-Income Countries: beyond the 'best BUYS'

   International Journal of Health Policy and Management 10:697–705.

   https://doi.org/10.34172/ijhpm.2020.245

   • PubMed
   • Google Scholar
2. 1. Bourhy H
   2. Nakouné E
   3. Hall M
   4. Nouvellet P
   5. Lepelletier A
   6. Talbi C
   7. Watier L
   8. Holmes EC
   9. Cauchemez S
   10. Lemey P
   11. Donnelly CA
   12. Rambaut A
   13. Parrish C

   (2016) Revealing the micro-scale signature of endemic zoonotic disease transmission in an African urban setting

   PLOS Pathogens 12:e1005525.

   https://doi.org/10.1371/journal.ppat.1005525

   • PubMed
   • Google Scholar
3. 1. Brunker K
   2. Marston DA
   3. Horton DL
   4. Cleaveland S
   5. Fooks AR
   6. Kazwala R
   7. Ngeleja C
   8. Lembo T
   9. Sambo M
   10. Mtema ZJ
   11. Sikana L
   12. Wilkie G
   13. Biek R
   14. Hampson K

   (2015) Elucidating the phylodynamics of endemic rabies virus in eastern Africa using whole-genome sequencing

   Virus Evolution 1:vev011.

   https://doi.org/10.1093/ve/vev011

   • PubMed
   • Google Scholar
4. 1. Brunker K
   2. Jaswant G
   3. Thumbi SM
   4. Lushasi K
   5. Lugelo A
   6. Czupryna AM
   7. Ade F
   8. Wambura G
   9. Chuchu V
   10. Steenson R
   11. Ngeleja C
   12. Bautista C
   13. Manalo DL
   14. Gomez MRR
   15. Chu M
   16. Miranda ME
   17. Kamat M
   18. Rysava K
   19. Espineda J
   20. Silo EAV
   21. Aringo AM
   22. Bernales RP
   23. Adonay FF
   24. Tildesley MJ
   25. Marston DA
   26. Jennings DL
   27. Fooks AR
   28. Zhu W
   29. Meredith LW
   30. Hill SC
   31. Poplawski R
   32. Gifford RJ
   33. Singer JB
   34. Maturi M
   35. Mwatondo A
   36. Biek R
   37. Hampson K

   (2020) Rapid in-country sequencing of whole virus genomes to inform rabies elimination programmes

   Wellcome Open Research 5:3.

   https://doi.org/10.12688/wellcomeopenres.15518.2

   • PubMed
   • Google Scholar
5. 1. Campbell K
   2. Gifford RJ
   3. Singer J
   4. Hill V
   5. O’Toole A
   6. Rambaut A
   7. Hampson K
   8. Brunker K

   (2022) Making genomic surveillance deliver: a lineage classification and nomenclature system to inform Rabies elimination

   PLOS Pathogens 18:e1010023.

   https://doi.org/10.1371/journal.ppat.1010023

   • PubMed
   • Google Scholar
6. 1. Castillo-Neyra R
   2. Brown J
   3. Borrini K
   4. Arevalo C
   5. Levy MZ
   6. Buttenheim A
   7. Hunter GC
   8. Becerra V
   9. Behrman J
   10. Paz-Soldan VA

   (2017) Barriers to dog Rabies vaccination during an urban Rabies outbreak: qualitative findings from Arequipa, Peru

   PLOS Neglected Tropical Diseases 11:e0005460.

   https://doi.org/10.1371/journal.pntd.0005460

   • PubMed
   • Google Scholar
7. 1. Changalucha J
   2. Steenson R
   3. Grieve E
   4. Cleaveland S
   5. Lembo T
   6. Lushasi K
   7. Mchau G
   8. Mtema Z
   9. Sambo M
   10. Nanai A
   11. Govella NJ
   12. Dilip A
   13. Sikana L
   14. Ventura F
   15. Hampson K

   (2019) The need to improve access to Rabies post-exposure vaccines: lessons from Tanzania

   Vaccine 37 Suppl 1:A45–A53.

   https://doi.org/10.1016/j.vaccine.2018.08.086

   • PubMed
   • Google Scholar
8. 1. Cleaveland S
   2. Kaare M
   3. Tiringa P
   4. Mlengeya T
   5. Barrat J

   (2003) A dog Rabies vaccination campaign in rural Africa: impact on the incidence of dog Rabies and human dog-bite injuries

   Vaccine 21:1965–1973.

   https://doi.org/10.1016/s0264-410x(02)00778-8

   • PubMed
   • Google Scholar
9. 1. Coetzer A
   2. Gwenhure L
   3. Makaya P
   4. Markotter W
   5. Nel L

   (2019) Epidemiological aspects of the persistent transmission of Rabies during an outbreak (2010 – 2017) in Harare, Zimbabwe

   PLOS ONE 14:e0210018.

   https://doi.org/10.1371/journal.pone.0210018

   • PubMed
   • Google Scholar
10. 1. Collective french multidisciplinary investigation team

    (2008) Identification of a rabid dog in France illegally introduced from Morocco

    Eurosurveillance 13:1–2.

    https://doi.org/10.2807/ese.13.11.08066-en

    • Google Scholar
11. 1. Cori A
    2. Nouvellet P
    3. Garske T
    4. Bourhy H
    5. Nakouné E
    6. Jombart T

    (2018) A graph-based evidence synthesis approach to detecting outbreak clusters: an application to dog Rabies

    PLOS Computational Biology 14:e1006554.

    https://doi.org/10.1371/journal.pcbi.1006554

    • PubMed
    • Google Scholar
12. 1. Czupryna AM
    2. Brown JS
    3. Bigambo MA
    4. Whelan CJ
    5. Mehta SD
    6. Santymire RM
    7. Lankester FJ
    8. Faust LJ

    (2016) Ecology and demography of free-roaming domestic dogs in rural villages near Serengeti National Park in Tanzania

    PLOS ONE 11:e0167092.

    https://doi.org/10.1371/journal.pone.0167092

    • PubMed
    • Google Scholar
13. 1. Davlin SL
    2. Vonville HM

    (2012) Canine Rabies vaccination and domestic dog population characteristics in the developing world: a systematic review

    Vaccine 30:3492–3502.

    https://doi.org/10.1016/j.vaccine.2012.03.069

    • PubMed
    • Google Scholar
14. 1. Faye O
    2. Boëlle P-Y
    3. Heleze E
    4. Faye O
    5. Loucoubar C
    6. Magassouba N
    7. Soropogui B
    8. Keita S
    9. Gakou T
    10. Bah EHI
    11. Koivogui L
    12. Sall AA
    13. Cauchemez S

    (2015) Chains of transmission and control of Ebola virus disease in Conakry, Guinea, in 2014: an observational study

    The Lancet Infectious Diseases 15:320–326.

    https://doi.org/10.1016/S1473-3099(14)71075-8

    • PubMed
    • Google Scholar
15. 1. Hampson K
    2. Coudeville L
    3. Lembo T
    4. Sambo M
    5. Kieffer A
    6. Attlan M
    7. Barrat J
    8. Blanton JD
    9. Briggs DJ
    10. Cleaveland S
    11. Costa P
    12. Freuling CM
    13. Hiby E
    14. Knopf L
    15. Leanes F
    16. Meslin F-X
    17. Metlin A
    18. Miranda ME
    19. Müller T
    20. Nel LH
    21. Recuenco S
    22. Rupprecht CE
    23. Schumacher C
    24. Taylor L
    25. Vigilato MAN
    26. Zinsstag J
    27. Dushoff J
    28. on behalf of the Global Alliance for Rabies Control Partners for Rabies Prevention

    (2015) Estimating the global burden of Endemic canine Rabies

    PLOS Neglected Tropical Diseases 9:e0003709.

    https://doi.org/10.1371/journal.pntd.0003709

    • Google Scholar
16. Preprint

    1. Hampson K
    2. Abela-Ridder B
    3. Brunker K
    4. Bucheli STM
    5. Carvalho M
    6. Caldas E
    7. Changalucha J
    8. Cleaveland S
    9. Dushoff J
    10. Gutierrez V
    11. Fooks AR
    12. Hotopp K
    13. Haydon DT
    14. Lugelo A
    15. Lushasi K
    16. Mancy R
    17. Marston DA
    18. Mtema Z
    19. Rajeev M
    20. Dourado L
    21. Gonzalez Roldan JF
    22. Rysava K
    23. Rocha SM
    24. Sambo M
    25. Sikana L
    26. Vigilato M
    27. Del Rio Vilas V

    (2016) Surveillance to Establish Elimination of Transmission and Freedom from Dog-Mediated Rabies

    bioRxiv.

    https://doi.org/10.1101/096883

    • Google Scholar
17. 1. Hampson K
    2. Ventura F
    3. Steenson R
    4. Mancy R
    5. Trotter C
    6. Cooper L
    7. Abela-Ridder B
    8. Knopf L
    9. Ringenier M
    10. Tenzin T
    11. Ly S
    12. Tarantola A
    13. Moyengar R
    14. Oussiguéré A
    15. Bonfoh B
    16. Narayana DA
    17. Sudarshan MK
    18. Muturi M
    19. Mwatondo A
    20. Wambura G
    21. Andriamandimby SF
    22. Baril L
    23. Edosoa GT
    24. Traoré A
    25. Jayme S
    26. Kotzé J
    27. Gunesekera A
    28. Chitnis N
    29. Hattendorf J
    30. Laager M
    31. Lechenne M
    32. Zinsstag J
    33. Changalucha J
    34. Mtema Z
    35. Lugelo A
    36. Lushasi K
    37. Yurachai O
    38. Metcalf CJE
    39. Rajeev M
    40. Blanton J
    41. Costa GB
    42. Sreenivasan N
    43. Wallace R
    44. Briggs D
    45. Taylor L
    46. Thumbi SM
    47. Huong NTT

    (2019) The potential effect of improved provision of Rabies post-exposure prophylaxis in Gavi-eligible countries: a Modelling study

    The Lancet Infectious Diseases 19:102–111.

    https://doi.org/10.1016/S1473-3099(18)30512-7

    • Google Scholar
18. 1. Jombart T

    (2008) Adegenet: a R package for the multivariate analysis of genetic markers

    Bioinformatics 24:1403–1405.

    https://doi.org/10.1093/bioinformatics/btn129

    • PubMed
    • Google Scholar
19. 1. Jombart T
    2. Ahmed I

    (2011) Adegenet 1.3-1: new tools for the analysis of genome-wide SNP data

    Bioinformatics 27:3070–3071.

    https://doi.org/10.1093/bioinformatics/btr521

    • PubMed
    • Google Scholar
20. 1. Kunkel A
    2. Jeon S
    3. Joseph HC
    4. Dilius P
    5. Crowdis K
    6. Meltzer MI
    7. Wallace R

    (2021) The urgency of resuming disrupted dog Rabies vaccination campaigns: a modeling and cost-effectiveness analysis

    Scientific Reports 11:12476.

    https://doi.org/10.1038/s41598-021-92067-5

    • PubMed
    • Google Scholar
21. 1. Lushasi K
    2. Hayes S
    3. Ferguson EA
    4. Changalucha J
    5. Cleaveland S
    6. Govella NJ
    7. Haydon DT
    8. Sambo M
    9. Mchau GJ
    10. Mpolya EA
    11. Mtema Z
    12. Nonga HE
    13. Steenson R
    14. Nouvellet P
    15. Donnelly CA
    16. Hampson K

    (2021) Reservoir dynamics of Rabies in South‐East Tanzania and the roles of cross‐species transmission and domestic dog vaccination

    The Journal of Applied Ecology 58:2673–2685.

    https://doi.org/10.1111/1365-2664.13983

    • PubMed
    • Google Scholar
22. 1. Mancy R
    2. Rajeev M
    3. Lugelo A
    4. Brunker K
    5. Cleaveland S
    6. Ferguson EA
    7. Hotopp K
    8. Kazwala R
    9. Magoto M
    10. Rysava K
    11. Haydon DT
    12. Hampson K

    (2022) Rabies shows how scale of transmission can enable acute infections to persist at low prevalence

    Science 376:512–516.

    https://doi.org/10.1126/science.abn0713

    • PubMed
    • Google Scholar
23. 1. Marston DA
    2. Jennings DL
    3. MacLaren NC
    4. Dorey-Robinson D
    5. Fooks AR
    6. Banyard AC
    7. McElhinney LM

    (2019) Pan-Lyssavirus real time RT-PCR for Rabies diagnosis

    Journal of Visualized Experiments 1:59709.

    https://doi.org/10.3791/59709

    • PubMed
    • Google Scholar
24. 1. Minghui R
    2. Stone M
    3. Semedo MH
    4. Nel L

    (2018) New global strategic plan to eliminate dog-mediated Rabies by 2030

    The Lancet Global Health 6:e828–e829.

    https://doi.org/10.1016/S2214-109X(18)30302-4

    • PubMed
    • Google Scholar
25. 1. Mpolya EA
    2. Lembo T
    3. Lushasi K
    4. Mancy R
    5. Mbunda EM
    6. Makungu S
    7. Maziku M
    8. Sikana L
    9. Jaswant G
    10. Townsend S
    11. Meslin F-X
    12. Abela-Ridder B
    13. Ngeleja C
    14. Changalucha J
    15. Mtema Z
    16. Sambo M
    17. Mchau G
    18. Rysava K
    19. Nanai A
    20. Kazwala R
    21. Cleaveland S
    22. Hampson K

    (2017) Toward elimination of dog-mediated human Rabies: experiences from implementing a large-scale demonstration project in Southern Tanzania

    Frontiers in Veterinary Science 4:21.

    https://doi.org/10.3389/fvets.2017.00021

    • PubMed
    • Google Scholar
26. 1. Nadal D
    2. Beeching S
    3. Cleaveland S
    4. Cronin K
    5. Hampson K
    6. Steenson R
    7. Abela-Ridder B

    (2022) Rabies and the pandemic: lessons for One Health

    Transactions of the Royal Society of Tropical Medicine and Hygiene 116:197–200.

    https://doi.org/10.1093/trstmh/trab123

    • PubMed
    • Google Scholar
27. 1. Nakamura T
    2. Yamada KD
    3. Tomii K
    4. Katoh K

    (2018) Parallelization of MAFFT for large-scale multiple sequence alignments

    Bioinformatics 34:2490–2492.

    https://doi.org/10.1093/bioinformatics/bty121

    • PubMed
    • Google Scholar
28. Report

    1. National Bureau of Statistics Tanzania

    (2012) Population and housing census

    NBS.

    https://www.nbs.go.tz/index.php/en/census-surveys/population-and-housing-census

    • Google Scholar
29. 1. Nel LH

    (2013) Discrepancies in data reporting for Rabies, Africa

    Emerging Infectious Diseases 19:529–533.

    https://doi.org/10.3201/eid1904.120185

    • PubMed
    • Google Scholar
30. 1. Nguyen L-T
    2. Schmidt HA
    3. von Haeseler A
    4. Minh BQ

    (2015) IQ-TREE: a fast and effective stochastic algorithm for estimating maximum-likelihood phylogenies

    Molecular Biology and Evolution 32:268–274.

    https://doi.org/10.1093/molbev/msu300

    • Google Scholar
31. 1. Paradis E
    2. Schliep K

    (2019) Ape 5.0: an environment for modern phylogenetics and evolutionary analyses in R

    Bioinformatics 35:526–528.

    https://doi.org/10.1093/bioinformatics/bty633

    • PubMed
    • Google Scholar
32. 1. Quick J
    2. Loman NJ
    3. Duraffour S
    4. Simpson JT
    5. Severi E
    6. Cowley L
    7. Bore JA
    8. Koundouno R
    9. Dudas G
    10. Mikhail A
    11. Ouédraogo N
    12. Afrough B
    13. Bah A
    14. Baum JH
    15. Becker-Ziaja B
    16. Boettcher J-P
    17. Cabeza-Cabrerizo M
    18. Camino-Sanchez A
    19. Carter LL
    20. Doerrbecker J
    21. Enkirch T
    22. Dorival IGG
    23. Hetzelt N
    24. Hinzmann J
    25. Holm T
    26. Kafetzopoulou LE
    27. Koropogui M
    28. Kosgey A
    29. Kuisma E
    30. Logue CH
    31. Mazzarelli A
    32. Meisel S
    33. Mertens M
    34. Michel J
    35. Ngabo D
    36. Nitzsche K
    37. Pallash E
    38. Patrono LV
    39. Portmann J
    40. Repits JG
    41. Rickett NY
    42. Sachse A
    43. Singethan K
    44. Vitoriano I
    45. Yemanaberhan RL
    46. Zekeng EG
    47. Trina R
    48. Bello A
    49. Sall AA
    50. Faye O
    51. Faye O
    52. Magassouba N
    53. Williams CV
    54. Amburgey V
    55. Winona L
    56. Davis E
    57. Gerlach J
    58. Washington F
    59. Monteil V
    60. Jourdain M
    61. Bererd M
    62. Camara A
    63. Somlare H
    64. Camara A
    65. Gerard M
    66. Bado G
    67. Baillet B
    68. Delaune D
    69. Nebie KY
    70. Diarra A
    71. Savane Y
    72. Pallawo RB
    73. Gutierrez GJ
    74. Milhano N
    75. Roger I
    76. Williams CJ
    77. Yattara F
    78. Lewandowski K
    79. Taylor J
    80. Rachwal P
    81. Turner D
    82. Pollakis G
    83. Hiscox JA
    84. Matthews DA
    85. O’Shea MK
    86. Johnston AM
    87. Wilson D
    88. Hutley E
    89. Smit E
    90. Di Caro A
    91. Woelfel R
    92. Stoecker K
    93. Fleischmann E
    94. Gabriel M
    95. Weller SA
    96. Koivogui L
    97. Diallo B
    98. Keita S
    99. Rambaut A
    100. Formenty P
    101. Gunther S
    102. Carroll MW

    (2016) Real-time, portable genome sequencing for Ebola surveillance

    Nature 530:228–232.

    https://doi.org/10.1038/nature16996

    • PubMed
    • Google Scholar
33. Software

    1. Rajeev M

    (2023) Treerabid, version 864f671

    GitHub.

    https://github.com/mrajeev08/treerabid
34. 1. Rambaut A
    2. Lam TT
    3. Max Carvalho L
    4. Pybus OG

    (2016) Exploring the temporal structure of heterochronous sequences using Tempest (formerly path-O-Gen)

    Virus Evolution 2:vew007.

    https://doi.org/10.1093/ve/vew007

    • Google Scholar
35. 1. Raynor B
    2. Díaz EW
    3. Shinnick J
    4. Zegarra E
    5. Monroy Y
    6. Mena C
    7. De la Puente-León M
    8. Levy MZ
    9. Castillo-Neyra R

    (2021) The impact of the COVID-19 pandemic on Rabies reemergence in Latin America: the case of Arequipa, Peru

    PLOS Neglected Tropical Diseases 15:e0009414.

    https://doi.org/10.1371/journal.pntd.0009414

    • PubMed
    • Google Scholar
36. Book

    1. Rupprecht CE
    2. Fooks AR
    3. Abela-Ridder B

    (2018)

    Laboratory techniques in rabies, volume 1, 5th ed.

    World Health Organization.

    • Google Scholar
37. 1. Rysava K
    2. Mancero T
    3. Caldas E
    4. de Carvalho MF
    5. Castro APB
    6. Gutiérrez V
    7. Haydon DT
    8. Johnson PCD
    9. Mancy R
    10. Montebello LR
    11. Rocha SM
    12. Gonzalez Roldan JF
    13. Vigilato MAN
    14. Vilas VDR
    15. Hampson K

    (2020) Towards the elimination of dog-mediated Rabies: development and application of an evidence-based management tool

    BMC Infectious Diseases 20:778.

    https://doi.org/10.1186/s12879-020-05457-x

    • PubMed
    • Google Scholar
38. 1. Sabeta C
    2. Ngoepe EC

    (2018) Controlling dog Rabies in Africa: successes, failures and prospects for the future

    Revue Scientifique et Technique 37:439–449.

    https://doi.org/10.20506/rst.37.2.2813

    • PubMed
    • Google Scholar
39. 1. Sambo M
    2. Johnson PCD
    3. Hotopp K
    4. Changalucha J
    5. Cleaveland S
    6. Kazwala R
    7. Lembo T
    8. Lugelo A
    9. Lushasi K
    10. Maziku M
    11. Mbunda E
    12. Mtema Z
    13. Sikana L
    14. Townsend SE
    15. Hampson K

    (2017) Comparing methods of assessing dog Rabies vaccination coverage in rural and urban communities in Tanzania

    Frontiers in Veterinary Science 4:33.

    https://doi.org/10.3389/fvets.2017.00033

    • PubMed
    • Google Scholar
40. 1. Sambo M
    2. Hampson K
    3. Changalucha J
    4. Cleaveland S
    5. Lembo T
    6. Lushasi K
    7. Mbunda E
    8. Mtema Z
    9. Sikana L
    10. Johnson PCD

    (2018) Estimating the size of dog populations in Tanzania to inform Rabies control

    Veterinary Sciences 5:5.

    https://doi.org/10.3390/vetsci5030077

    • PubMed
    • Google Scholar
41. 1. Sharma B
    2. Dhand NK
    3. Timsina N
    4. Ward MP

    (2010) Reemergence of rabies in Chhukha district, Bhutan, 2008

    Emerging Infectious Diseases 16:1925–1930.

    https://doi.org/10.3201/eid1612.100958

    • PubMed
    • Google Scholar
42. 1. Suchard MA
    2. Lemey P
    3. Baele G
    4. Ayres DL
    5. Drummond AJ
    6. Rambaut A

    (2018) Bayesian phylogenetic and phylodynamic data integration using BEAST 1.10

    Virus Evolution 4:vey016.

    https://doi.org/10.1093/ve/vey016

    • PubMed
    • Google Scholar
43. 1. Swedberg C
    2. Mazeri S
    3. Mellanby RJ
    4. Hampson K
    5. Chng NR

    (2022) Implementing a one health approach to Rabies surveillance: lessons from integrated bite case management

    Frontiers in Tropical Diseases 3:829132.

    https://doi.org/10.3389/fitd.2022.829132

    • PubMed
    • Google Scholar
44. 1. Tohma K
    2. Saito M
    3. Demetria CS
    4. Manalo DL
    5. Quiambao BP
    6. Kamigaki T
    7. Oshitani H

    (2016) Molecular and mathematical modeling analyses of inter-Island transmission of Rabies into a previously Rabies-free Island in the Philippines

    Infection, Genetics and Evolution 38:22–28.

    https://doi.org/10.1016/j.meegid.2015.12.001

    • PubMed
    • Google Scholar
45. 1. Townsend SE
    2. Lembo T
    3. Cleaveland S
    4. Meslin FX
    5. Miranda ME
    6. Putra AAG
    7. Haydon DT
    8. Hampson K

    (2013a) Surveillance guidelines for disease elimination: a case study of canine Rabies

    Comparative Immunology, Microbiology and Infectious Diseases 36:249–261.

    https://doi.org/10.1016/j.cimid.2012.10.008

    • PubMed
    • Google Scholar
46. 1. Townsend SE
    2. Sumantra IP
    3. Bagus GN
    4. Brum E
    5. Cleaveland S
    6. Crafter S
    7. Dewi APM
    8. Dharma DMN
    9. Dushoff J
    10. Girardi J
    11. Gunata IK
    12. Hiby EF
    13. Kalalo C
    14. Knobel DL
    15. Mardiana IW
    16. Putra AAG
    17. Schoonman L
    18. Scott-Orr H
    19. Shand M
    20. Sukanadi IW
    21. Suseno PP
    22. Haydon DT
    23. Hampson K

    (2013b) Designing programs for eliminating canine Rabies from Islands: Bali, Indonesia as a case study

    PLOS Neglected Tropical Diseases 7:e2372.

    https://doi.org/10.1371/journal.pntd.0002372

    • PubMed
    • Google Scholar
47. 1. Vigilato MAN
    2. Clavijo A
    3. Knobl T
    4. Silva HMT
    5. Cosivi O
    6. Schneider MC
    7. Leanes LF
    8. Belotto AJ
    9. Espinal MA

    (2013) Progress towards eliminating canine Rabies: policies and perspectives from Latin America and the Caribbean

    Philosophical Transactions of the Royal Society B 368:20120143.

    https://doi.org/10.1098/rstb.2012.0143

    • Google Scholar
48. 1. Windiyaningsih C
    2. Wilde H
    3. Meslin FX
    4. Suroso T
    5. Widarso H

    (2004)

    The rabies epidemic on Flores Island, Indonesia (1998-2003)

    J Med Assoc Thail Chotmaihet Thangphaet 87:1389–1393.

    • Google Scholar
49. Report

    1. World Health Organization

    (2018a)

    WHO Expert Consultation on Rabies, third report (WHO Technical Report Series, No. 1012)

    Geneva, Switzerland: World Health Organization.

    • Google Scholar
50. 1. World Health Organization

    (2018b) Rabies vaccines: WHO position paper, April 2018 - Recommendations

    Vaccine 1:36.

    https://doi.org/10.1016/j.vaccine.2018.06.061

    • PubMed
    • Google Scholar
51. 1. Yu G

    (2020) Using Ggtree to visualize data on Tree‐Like structures

    Current Protocols in Bioinformatics 69:e96.

    https://doi.org/10.1002/cpbi.96

    • PubMed
    • Google Scholar
52. 1. Zinsstag J
    2. Lechenne M
    3. Laager M
    4. Mindekem R
    5. Naïssengar S
    6. Oussiguéré A
    7. Bidjeh K
    8. Rives G
    9. Tessier J
    10. Madjaninan S
    11. Ouagal M
    12. Moto DD
    13. Alfaroukh IO
    14. Muthiani Y
    15. Traoré A
    16. Hattendorf J
    17. Lepelletier A
    18. Kergoat L
    19. Bourhy H
    20. Dacheux L
    21. Stadler T
    22. Chitnis N

    (2017) Vaccination of dogs in an African city Interrupts Rabies transmission and reduces human exposure

    Science Translational Medicine 9:eaaf6984.

    https://doi.org/10.1126/scitranslmed.aaf6984

    • PubMed
    • Google Scholar

Decision letter after peer review:

Thank you for submitting your article "Integrating contact tracing and whole-genome sequencing to track the elimination of dog-mediated rabies: an observational and genomic study" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, including Jennifer Flegg as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by Miles Davenport as the Senior Editor.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

The following questions need to be addressed:

1) Can the effective reproduction number be estimated from a combination of genetic and contact tracing data? Or can the author comment on the possibility of integrating both types of data into the estimation process?

2) The health economics analysis is too preliminary to draw any strong conclusions from. I think more could be done to improve the health economics methods used.

3) I think the paper oversells a little from its findings. The paper uses a rudimentary health economics approach to make large claims about the cost-effectiveness of different approaches.

4) There has been an important effort to collect all the data presented in this paper. Some of the data are important to better understand transmission dynamics and may not be necessary to monitor and control rabies; some of the datasets may also be expensive to collect, which may be possible to do with the support of research teams but might be impossible to achieve in other settings and on wider areas. I think it would be interesting if authors could discuss how minimal additional data collection would improve the monitoring and control of rabies in an African setting and why accounting for the constraints on cost. I'd also be interested in extending the discussion about the difference between data collected for research (here) and data collected for surveillance.

Reviewer #1 (Recommendations for the authors):

1) Can the effective reproduction number be estimated from a combination of genetic and contact tracing data? Or can the authors comment on the possibility of integrating both types of data into the estimation process?

2) The health economics analysis is too preliminary to draw any strong conclusions from. This needs further comment or, ideally, further work.

3) I think the paper oversells a little from its findings. The paper uses a rudimentary health economics approach to make large claims about the cost-effectiveness of different approaches.

4) There has been an important effort to collect all the data presented in this paper. Some of the data are important to better understand transmission dynamics and may not be necessary to monitor and control rabies; some of the datasets may also be expansive to collect, which may be possible to do with the support of research teams but might be impossible to achieve in other settings and on wider areas. I think it would be interesting if authors could discuss how minimal additional data collection would improve the monitoring and control of rabies in an African setting and why accounting for the constraints on cost. I'd also be interested in extending the discussion about the difference between data collected for research (here) and data collected for surveillance.

5) The phylogenetic analyses were difficult to understand. The authors use a phylogenetic framework to estimate the underlying number of rabid dogs per outbreak (171 in the first outbreak and 140 in the second one), but it was unclear to me where the information was coming from. From the supplementary material, it seems the authors build transmission trees consistent with the phylogenies. However, these are reliant on (a) a serial interval and (b) a dispersal kernel. There is no reference as to what serial interval distribution was used and how it was calculated. Similarly, there is no information on the dispersal kernel, including what data was used to fit it. I suspect that the serial interval for rabies (and probably the dispersal kernel) has a long tail, which would lead to substantial uncertainty in the transmission chains, however, I could not see uncertainty in the outbreak sizes.

6) Relatedly, it seems the transmission chain reconstruction relies on identifying missing cases that link observed cases. However, it was unclear to me whether this approach would detect the presence of many dead ends (e.g., an over-dispersed offspring distribution with many cases that don't result in onward transmission) – I could imagine this would lead to similar phylogenetic trees (and probably dispersal kernels) but with a very different number of cases.

7) In Figure S4 panel B, there seems to be a quite even distribution in the probability of linking a pair consistently between 0 and 1 (although I may be incorrect in my interpretation of this figure). Does this not mean that there is ultimately little information in the data on the transmission chains? As these plots are a little confusing to interpret, it would be good to provide guidance as to how to interpret them. I would also provide a written assessment (in the main text) of the authors' ability to accurately reconstruct transmission trees.

8) Given the central importance of the outbreak sizes to the paper – including providing estimates of the case detection probabilities and onwards estimates of the economic impact of the vaccine policies, it would be good to understand the importance of correctly understanding the serial interval and the dispersal kernel to the eventual estimates of the total number of infections. The authors could run sensitivity analyses where these terms are varied.

9) Relatedly, I was not sure of how even case reporting is over space and over time on Pemba. Similarly, it was not clear how representative the sequenced viruses are in space and time (ie was there extra sequencing in some years/some parts of the island). I also wondered whether the fact that sequences come from contact tracing and are therefore not independent samples was problematic. I would set out any key assumptions (and potential limitations) with the phylogenetic approach.

Reviewer #3 (Recommendations for the authors):

Relatedly, it seems the transmission chain reconstruction relies on identifying missing cases that link observed cases. However, it was unclear to me whether this approach would detect the presence of many dead ends (e.g., an over-dispersed offspring distribution with many cases that don't result in onwards transmission) – I could imagine this would lead to similar phylogenetic trees (and probably dispersal kernels) but with a very different number of cases.

In Figure S4 panel B, there seems to be a quite even distribution in the probability of linking a pair consistently between 0 and 1 (although I may be incorrect in my interpretation of this figure). Does this not mean that there is ultimately little information in the data on the transmission chains? As these plots are a little confusing to interpret, it would be good to provide guidance as to how to interpret them. I would also provide a written assessment (in the main text) of the authors' ability to accurately reconstruct transmission trees.

Given the central importance of the outbreak sizes to the paper – including providing estimates of the case detection probabilities and onwards estimates of the economic impact of the vaccine policies, it would be good to understand the importance of correctly understanding the serial interval and the dispersal kernel to the eventual estimates of the total number of infections. The authors could run sensitivity analyses where these terms are varied.

Relatedly, I was not sure of how even case reporting is over space and over time on Pemba. Similarly, it was not clear how representative the sequenced viruses are in space and time (ie was there extra sequencing in some years/some parts of the island). I also wondered whether the fact that sequences come from contact tracing and are therefore not independent samples was problematic. I would set out any key assumptions (and potential limitations) with the phylogenetic approach.

Essential revisions:

The following questions need to be addressed:

1) Can the effective reproduction number be estimated from a combination of genetic and contact tracing data? Or can the author comment on the possibility of integrating both types of data into the estimation process?

We previously estimated the effective reproductive number (R_e) from our transmission tree reconstructions that incorporated both the viral whole genome sequences and contact tracing data. We have now updated Figure 2 with a smoothed LOESS regression of R_e over time to better visualise the estimate and uncertainty around it. We also moved the methods and results on R_e from the supplement to the main text and added some discussion about how the inclusion of both data sources affects our inference, and how future work could build on these approaches.

Specifically in the revision we now highlight that the genomic data reveals that two introductions occurred in 2016 whereas we would have naively expected the outbreak to be the result of a single introduction; and that no other introductions after 2016 established onward rabies transmission on Pemba. These analyses instil confidence that rabies was eliminated not just once but twice confirming the feasibility of elimination through dog vaccination, and reassuringly show that rabies was not circulating undetected from 2014 to 2016 (ruled out by the genomic data) and only re-emerged because of external introductions.

2) The health economics analysis is too preliminary to draw any strong conclusions from. I think more could be done to improve the health economics methods used.

In our revision we have drawn from previously published work to develop a decision tree model of the impact of rabies control and prevention measures (focusing on PEP and mass dog vaccination). We take parameters estimated, where possible, directly from the Pemba data or from the literature e.g. the probability of progression to rabies in the absence of PEP (which was estimated from mainland Tanzania contact tracing data so is expected to be robust in this setting also). Using this model we now project the cost-effectiveness of the different intervention components as part of counterfactual scenarios over a 10-year time horizon. We highlight that a key limitation of our health economics work is the degree to which the conclusions are transferable to other settings, so we also more strongly caveat our findings in the revised discussion and describe other limitations of this approach.

3) I think the paper oversells a little from its findings. The paper uses a rudimentary health economics approach to make large claims about the cost-effectiveness of different approaches.

We agree that our health economics approach was rather rudimentary (see point 2) and that it would be appropriate to be more cautious in generalising from our findings. In our revision we have undertaken a more robust cost-effectiveness analysis, and toned down our claims accordingly to focus on Pemba. We also highlight important uncertainties that should be considered for future work, that would increase the transferability of the approaches we describe.

4) There has been an important effort to collect all the data presented in this paper. Some of the data are important to better understand transmission dynamics and may not be necessary to monitor and control rabies; some of the datasets may also be expensive to collect, which may be possible to do with the support of research teams but might be impossible to achieve in other settings and on wider areas. I think it would be interesting if authors could discuss how minimal additional data collection would improve the monitoring and control of rabies in an African setting and why accounting for the constraints on cost. I'd also be interested in extending the discussion about the difference between data collected for research (here) and data collected for surveillance.

We really appreciate this comment. Our approach was thorough for purposes of research, and hence our data collection was more in-depth than routine government surveillance in Tanzania, which is extremely resource constrained. But, on reflection the contact tracing, outreach and sampling was much less in-depth than in high-income settings (for example in France: https://www.eurosurveillance.org/content/10.2807/ese.13.11.08066-en and the US https://www.forbes.com/sites/brucelee/2021/06/20/rabid-dog-imported-into-us-at-least-12-people-exposed/) and was one of the easier aspects of the study! Lead author KL carried out contact tracing (part-time) in liaison with four livestock officers all equipped with motorbikes for their routine work. However, investment to enable one highly-skilled person to focus on contact tracing rabies has not been considered a priority by LMIC governments or global health funders.

Similarly, development of sequencing protocols was a significant effort (led by KB), but the sequencing from Pemba was small-scale and going forward can be done routinely and cost-effectively by a skilled local scientist (for example, co-author GJ who has since trained several others). Bigger challenges were securing funds for interventions (dog vaccination and free PEP), for training of government animal and public health staff in these interventions and for Tanzanian scientists to undertake postgraduate degrees that led to this manuscript (KL, GJ, MS, LS)! With all this in mind we have clarified in the discussion (paragraph 4) how much can be achieved routinely and make recommendations as to how routine surveillance could be strengthened most efficiently drawing from our experiences.

Reviewer #1 (Recommendations for the authors):

1) Can the effective reproduction number be estimated from a combination of genetic and contact tracing data? Or can the authors comment on the possibility of integrating both types of data into the estimation process?

Yes! We previously estimated Re combining these data sources, but both the methodological details and results were in the supplement. We have now moved these to the main text. We also add some discussion about how the two sources of data are integrated and how this adds value to our inference.

2) The health economics analysis is too preliminary to draw any strong conclusions from. This needs further comment or, ideally, further work.

Agreed, we overstated our conclusions. In our revision we developed a more robust decision tree model so as to compare counterfactual scenarios (no intervention, the status quo where PEP is charged to patients, free PEP, and the combination of dog vaccination and free PEP) over a ten-year time horizon, taking the perspective of the health provider. We emphasise the degree to which these scenarios are idealised in comparison to the reality experienced on Pemba, the limitations of this approach and how future work could improve on this. We also point to the specific implications for Pemba and potential transferability of our conclusions to other settings.

3) I think the paper oversells a little from its findings. The paper uses a rudimentary health economics approach to make large claims about the cost-effectiveness of different approaches.

Agreed – see point 2.

4) There has been an important effort to collect all the data presented in this paper. Some of the data are important to better understand transmission dynamics and may not be necessary to monitor and control rabies; some of the datasets may also be expansive to collect, which may be possible to do with the support of research teams but might be impossible to achieve in other settings and on wider areas. I think it would be interesting if authors could discuss how minimal additional data collection would improve the monitoring and control of rabies in an African setting and why accounting for the constraints on cost. I'd also be interested in extending the discussion about the difference between data collected for research (here) and data collected for surveillance.

Thank you for this feedback. We have extended our discussion on these points.

5) The phylogenetic analyses were difficult to understand. The authors use a phylogenetic framework to estimate the underlying number of rabid dogs per outbreak (171 in the first outbreak and 140 in the second one), but it was unclear to me where the information was coming from. From the supplementary material, it seems the authors build transmission trees consistent with the phylogenies. However, these are reliant on (a) a serial interval and (b) a dispersal kernel. There is no reference as to what serial interval distribution was used and how it was calculated. Similarly, there is no information on the dispersal kernel, including what data was used to fit it. I suspect that the serial interval for rabies (and probably the dispersal kernel) has a long tail, which would lead to substantial uncertainty in the transmission chains, however, I could not see uncertainty in the outbreak sizes.

We apologise for these omissions. We have added detail on parameter estimates for the serial interval and dispersal kernel and confidence intervals around the estimated outbreak sizes. We also report our sensitivity analyses where we examine how robust our transmission chains are, and resulting metrics calculated from them, accounting for the long-tailed distributions of both the dispersal kernel and serial interval.

6) Relatedly, it seems the transmission chain reconstruction relies on identifying missing cases that link observed cases. However, it was unclear to me whether this approach would detect the presence of many dead ends (e.g., an over-dispersed offspring distribution with many cases that don't result in onward transmission) – I could imagine this would lead to similar phylogenetic trees (and probably dispersal kernels) but with a very different number of cases.

The reviewer is correct to point out how an overdispersed offspring distribution will lead to many deadends. This was a concern which we examined in some detail and now report in the main text. We were more likely to miss dead-end cases and short chains of transmission (see Figure 5 and supplement), but not sufficiently so to dramatically affect our outbreak estimates. From simulations we estimate that our method was more likely to have underestimated case detection, but not by more than 10%.

7) In Figure S4 panel B, there seems to be a quite even distribution in the probability of linking a pair consistently between 0 and 1 (although I may be incorrect in my interpretation of this figure). Does this not mean that there is ultimately little information in the data on the transmission chains? As these plots are a little confusing to interpret, it would be good to provide guidance as to how to interpret them. I would also provide a written assessment (in the main text) of the authors' ability to accurately reconstruct transmission trees.

These are great points. A limitation from the transmission reconstructions and integration with genomic data is that we cannot identify exactly who infected whom within transmission chains given the low mutation rate of rabies and short spatiotemporal distances between cases. However, we do have more confidence in the high-level topology of our transmission tree reconstructions and plausible progenitors within transmission chains (as most cases had only a few plausible progenitors) which we have explained now in the main text. Probably more important than exact transmission reconstruction is how the method allows us to estimate what proportion of cases from the outbreak were detected and thus determine bounds on the outbreak size. We have added some guidance on interpreting the transmission trees (Figure 4 supplement legends) and assessment of their accuracy in the main results.

8) Given the central importance of the outbreak sizes to the paper – including providing estimates of the case detection probabilities and onwards estimates of the economic impact of the vaccine policies, it would be good to understand the importance of correctly understanding the serial interval and the dispersal kernel to the eventual estimates of the total number of infections. The authors could run sensitivity analyses where these terms are varied.

This is an important point. In the revision we detail our parameter values (serial interval and dispersal kernel) which came from extensive contracting data elsewhere in Tanzania. We also report how robust our estimates of case detection were and thus the uncertainty in outbreak sizes (see point 7). From running various sensitivity analyses of the transmission tree reconstructions (including pruning at differing cuts offs for the serial interval and dispersal kernel distributions) we found these to have very little influence on overall case detection estimates. We explain this in the revision.

9) Relatedly, I was not sure of how even case reporting is over space and over time on Pemba. Similarly, it was not clear how representative the sequenced viruses are in space and time (ie was there extra sequencing in some years/some parts of the island). I also wondered whether the fact that sequences come from contact tracing and are therefore not independent samples was problematic. I would set out any key assumptions (and potential limitations) with the phylogenetic approach.

There was considerable variance in the degree to which cases were identified through contact tracing (typically from their case histories without recovery of a sample) and sequenced (samples were recovered and sequenced from 10% of identified cases, representing approximately 6% of cases estimated to have occurred). For instance, we inferred lower case detection in the initial endemic period (2010-2014, ~54%) than subsequently during the outbreak (2016-2018, 69%). There was no doubt further fine-scale variation in case detection across the island, which we plan to investigate in our continued work on integrating epidemiological and genomic data. However, we think that the work we present gives confidence as to how robust case detection from contact tracing is, with the majority of transmission chains exceeding 2 cases being detected. In our revision we present the robustness of case detection estimates in the main text, and lay out the key assumptions and potential limitations of the approaches that we used.

We do not think that the sequencing of cases identified through contact tracing present a problem in terms of lack of independence. We either combined the two data sources together to draw inference (transmission trees) or conducted the phylogenetic analyses using only the sequences and associated metadata to specifically address the question of introductions. We show in the sensitivity analyses that transmission trees inferred only from epidemiological data (without sequence data) were generally robust with respect to transmission chains (and to a lesser extent to plausible progenitors), but that pruning trees based on time cutoffs increased the congruence of transmission chains identified as consistent with the phylogenetic data. We have elaborated on these points in the revision.

Author details

1. Kennedy Lushasi

   1. Environmental Health and Ecological Sciences Department, Ifakara Health Institute, Dar es salaam, United Republic of Tanzania
   2. Boyd Orr Centre for Population and Ecosystem Health, School of Biodiversity, One Health & Veterinary Medicine, University of Glasgow, Glasgow, United Kingdom
   3. Department of Global Health and Biomedical Sciences, School of Life Sciences and Bioengineering, Nelson Mandela African Institution of Science and Technology, Arusha, United Republic of Tanzania

   Contribution

   Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing - original draft, Project administration

   Contributed equally with

   Kirstyn Brunker

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-2060-4202

2. Kirstyn Brunker

   Boyd Orr Centre for Population and Ecosystem Health, School of Biodiversity, One Health & Veterinary Medicine, University of Glasgow, Glasgow, United Kingdom

   Contribution

   Conceptualization, Data curation, Formal analysis, Supervision, Investigation, Visualization, Methodology, Writing - original draft, Project administration

   Contributed equally with

   Kennedy Lushasi

   Competing interests

   No competing interests declared

3. Malavika Rajeev

   Department of Ecology and Evolutionary Biology, Princeton University, Princeton, United States

   Contribution

   Formal analysis, Investigation, Visualization, Methodology

   Competing interests

   No competing interests declared

4. Elaine A Ferguson

   Boyd Orr Centre for Population and Ecosystem Health, School of Biodiversity, One Health & Veterinary Medicine, University of Glasgow, Glasgow, United Kingdom

   Contribution

   Formal analysis, Visualization

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-2010-765X

5. Gurdeep Jaswant

   1. Environmental Health and Ecological Sciences Department, Ifakara Health Institute, Dar es salaam, United Republic of Tanzania
   2. The University of Nairobi, Nairobi, Kenya

   Contribution

   Investigation

   Competing interests

   No competing interests declared

6. Laurie Louise Baker

   1. Boyd Orr Centre for Population and Ecosystem Health, School of Biodiversity, One Health & Veterinary Medicine, University of Glasgow, Glasgow, United Kingdom
   2. College of the Atlantic, Bar Harbor, United States

   Contribution

   Visualization

   Competing interests

   No competing interests declared

7. Roman Biek

   Boyd Orr Centre for Population and Ecosystem Health, School of Biodiversity, One Health & Veterinary Medicine, University of Glasgow, Glasgow, United Kingdom

   Contribution

   Supervision

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-3471-5357

8. Joel Changalucha

   1. Environmental Health and Ecological Sciences Department, Ifakara Health Institute, Dar es salaam, United Republic of Tanzania
   2. Boyd Orr Centre for Population and Ecosystem Health, School of Biodiversity, One Health & Veterinary Medicine, University of Glasgow, Glasgow, United Kingdom
   3. Animal & Plant Health Agency, Weybridge, United Kingdom

   Contribution

   Investigation

   Competing interests

   No competing interests declared

9. Sarah Cleaveland

   Boyd Orr Centre for Population and Ecosystem Health, School of Biodiversity, One Health & Veterinary Medicine, University of Glasgow, Glasgow, United Kingdom

   Contribution

   Funding acquisition, Writing - review and editing

   Competing interests

   No competing interests declared

10. Anna Czupryna

    Boyd Orr Centre for Population and Ecosystem Health, School of Biodiversity, One Health & Veterinary Medicine, University of Glasgow, Glasgow, United Kingdom

    Contribution

    Investigation

    Competing interests

    No competing interests declared

11. Anthony R Fooks

    Animal & Plant Health Agency, Weybridge, United Kingdom

    Contribution

    Resources, Funding acquisition

    Competing interests

    No competing interests declared

12. Nicodemus J Govella

    Environmental Health and Ecological Sciences Department, Ifakara Health Institute, Dar es salaam, United Republic of Tanzania

    Contribution

    Supervision, Project administration, Writing - review and editing

    Competing interests

    No competing interests declared

13. Daniel T Haydon

    Boyd Orr Centre for Population and Ecosystem Health, School of Biodiversity, One Health & Veterinary Medicine, University of Glasgow, Glasgow, United Kingdom

    Contribution

    Supervision, Writing - review and editing

    Competing interests

    No competing interests declared

14. Paul CD Johnson

    Boyd Orr Centre for Population and Ecosystem Health, School of Biodiversity, One Health & Veterinary Medicine, University of Glasgow, Glasgow, United Kingdom

    Contribution

    Formal analysis, Supervision

    Competing interests

    No competing interests declared

    "This ORCID iD identifies the author of this article:" 0000-0001-6663-7520

15. Rudovick Kazwala

    Department of Veterinary Medicine and Public Health, Sokoine University of Agriculture, Morogoro, United Republic of Tanzania

    Contribution

    Supervision, Funding acquisition, Project administration

    Competing interests

    No competing interests declared

16. Tiziana Lembo

    Boyd Orr Centre for Population and Ecosystem Health, School of Biodiversity, One Health & Veterinary Medicine, University of Glasgow, Glasgow, United Kingdom

    Contribution

    Funding acquisition, Investigation, Writing - review and editing

    Competing interests

    No competing interests declared

17. Denise Marston

    1. Animal & Plant Health Agency, Weybridge, United Kingdom
    2. Department of Comparative Biomedical Sciences, School of Veterinary Medicine, University of Surrey, Guildford, Surrey, United Kingdom

    Contribution

    Resources, Funding acquisition, Investigation, Writing - review and editing

    Competing interests

    No competing interests declared

18. Msanif Masoud

    Ministry for Health and Social Welfare, Zanzibar, United Republic of Tanzania

    Contribution

    Investigation

    Competing interests

    No competing interests declared

19. Matthew Maziku

    Ministry of Livestock Development and Fisheries, Dodoma, United Republic of Tanzania

    Contribution

    Investigation, Project administration

    Competing interests

    No competing interests declared

20. Eberhard Mbunda

    Ministry of Livestock Development and Fisheries, Dodoma, United Republic of Tanzania

    Contribution

    Investigation, Project administration

    Competing interests

    No competing interests declared

21. Geofrey Mchau

    Department of Epidemiology, Ministry of Health, Community Development, Gender, Elderly and Children (MoHCDGEC), Dodoma, United Republic of Tanzania

    Contribution

    Investigation, Project administration

    Competing interests

    No competing interests declared

22. Ally Z Mohamed

    Department of Livestock Development, Pemba, Ministry of Livestock Development and Fisheries, Zanzibar, United Republic of Tanzania

    Contribution

    Investigation

    Competing interests

    No competing interests declared

23. Emmanuel Mpolya

    Department of Global Health and Biomedical Sciences, School of Life Sciences and Bioengineering, Nelson Mandela African Institution of Science and Technology, Arusha, United Republic of Tanzania

    Contribution

    Supervision, Investigation

    Competing interests

    No competing interests declared

    "This ORCID iD identifies the author of this article:" 0000-0002-6210-9445

24. Chanasa Ngeleja

    Tanzania Veterinary Laboratory Agency, Dar es salaam, United Republic of Tanzania

    Contribution

    Resources, Investigation

    Competing interests

    No competing interests declared

25. Kija Ng'habi

    Mbeya college of Health and Allied Sciences, University of Dar es Salaam, Mbeya, United Republic of Tanzania

    Contribution

    Supervision

    Competing interests

    No competing interests declared

26. Hezron Nonga

    Ministry of Livestock Development and Fisheries, Dodoma, United Republic of Tanzania

    Contribution

    Resources, Investigation

    Competing interests

    No competing interests declared

27. Kassim Omar

    Department of Livestock Development, Pemba, Ministry of Livestock Development and Fisheries, Zanzibar, United Republic of Tanzania

    Contribution

    Investigation

    Competing interests

    No competing interests declared

28. Kristyna Rysava

    University of Warwick, Coventry, United Kingdom

    Contribution

    Investigation, Visualization

    Competing interests

    No competing interests declared

29. Maganga Sambo

    Environmental Health and Ecological Sciences Department, Ifakara Health Institute, Dar es salaam, United Republic of Tanzania

    Contribution

    Investigation

    Competing interests

    No competing interests declared

30. Lwitiko Sikana

    Environmental Health and Ecological Sciences Department, Ifakara Health Institute, Dar es salaam, United Republic of Tanzania

    Contribution

    Investigation

    Competing interests

    No competing interests declared

31. Rachel Steenson

    Boyd Orr Centre for Population and Ecosystem Health, School of Biodiversity, One Health & Veterinary Medicine, University of Glasgow, Glasgow, United Kingdom

    Contribution

    Formal analysis, Visualization

    Competing interests

    No competing interests declared

32. Katie Hampson

    Boyd Orr Centre for Population and Ecosystem Health, School of Biodiversity, One Health & Veterinary Medicine, University of Glasgow, Glasgow, United Kingdom

    Contribution

    Conceptualization, Resources, Data curation, Formal analysis, Supervision, Funding acquisition, Investigation, Methodology, Writing - original draft, Project administration

    For correspondence

    katie.hampson@glasgow.ac.uk

    Competing interests

    No competing interests declared

    "This ORCID iD identifies the author of this article:" 0000-0001-5392-6884

• 1,038

  Page views

• 150

  Downloads

• 0

  Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.