Key findings
Guidance for established rabies elimination programmes, such as those in Latin America, is relatively limited and not geographically specific (10, 29). The management tool that we have developed and validated using surveillance data from Mexico and Brazil allows to identify mutually exclusive epidemiological situations from a simple algorithm, without the need for extensive statistical expertise (illustrated at: https://boydorr.shinyapps.io/paho_rabies/). By classifying states in these countries, we determined surveillance and control priorities at local, national and regional levels, and derived tailored guidance on how to continue progressing towards elimination, while revealing insights into rabies dynamics (Figures 3 and 4, Box 2).
Progress towards elimination was evident in both Mexico and Brazil; however, focal transmission remains a threat for re-emergence in ostensibly rabies-free states. Connectivity has been demonstrated to play a critical role in rabies persistence, with infection maintained across, and driven by, large interconnected metapopulations (rather than dense conurbations) (30, 31). Our classification identified metapopulations that support focal transmission, such as in central Mexico in the recent past (e.g. in 2010, Figure 3), and the cross-border area between Mexico and Guatemala, as well as in Northeast Brazil (Figure 3). There is an urgent need to improve surveillance and control in remaining foci, including endemic bordering countries. For example, the outbreak in the municipality of Corumba, Mato Grosso do Sol, on the Brazil-Bolivia border is rumoured to have spread from Bolivia, even though no cases were reported to SIRVERA from Bolivia in over three years. Incursions with long-lasting consequences have been reported elsewhere in Latin America (32), and globally (19, 33), and threaten the regional elimination effort. Moreover, rabies still persists in a few states where control programmes have been ongoing for decades. We suggest that in these Endemic states, post-vaccination monitoring is needed to identify the causes of slow progress, which likely relate to inadequacies in dog vaccination campaigns. Vaccination coverage in Latin America has generally been estimated retrospectively from human:dog ratios. However, human:dog ratios can be heterogeneous and change considerably over time (34). Estimating numbers of distinct dogs vaccinated can also be difficult, as some dogs are vaccinated repeatedly during outbreak responses unless concerted efforts are made to target areas missed during campaigns. Gaps in coverage were previously found to be a critical factor in rabies control, prolonging progress towards elimination (19, 25). Post-campaign assessments of coverage are needed to both identify and remedy problematic areas (35). Strengthening the delivery and monitoring of dog vaccination campaigns in areas with focal transmission (including in neighbouring countries), is likely to be the single most important programmatic change for improving elimination prospects regionally.
Strengths and limitations
Surveillance quality affects the utility of evaluation tools to inform management. We considered surveillance quality to be sufficient throughout Mexico and Brazil based on assessment of indicators from the last 5 years including annual dog vaccinations in every state, adequate sample submissions and regular proficiency testing of laboratories (36), further supported by interviews with rabies programme managers and through PAHO missions. For other countries with less progressed control programmes and weaker surveillance, application of this tool would not be appropriate. For example, settings where locally-acquired human rabies cases are reported in the absence of confirmed animal cases would be indicative of inadequate surveillance. A particular concern is that the absence of detected cases reflects weak surveillance and not elimination. Although our requirement for a decline in case detection concomitant with dog vaccination prior to classification to putatively Absent guards against this, we recommend initial review to determine whether a country would benefit from this tool, for example, progression on SARE to at least step 3. Most countries in sub-Saharan Africa and Asia are probably at too early a stage in their control efforts for this tool to yet be of use, but we anticipate that in Latin America this tool could be applied to most countries and its use would complement SIRVERA.
Using sub-sampled simulated data, we demonstrated the extent to which our tool was robust to surveillance quality in endemic settings (Figure 2). Surveillance had to reach very low levels or be very biased for states to be misclassified (<2.5% detection in Endemic/Declining classifications). This threshold detection level increased to 5% as incursions become the main source of cases i.e. in Intermittent and Absent-Vulnerable classifications. We therefore emphasize the need to enhance surveillance to detect at least 5% of circulating cases, a recommendation consistent with prior work showing that 5% case detection is required to verify disease freedom (9). Integrated Bite Case Management shows promise as a tool that can detect >10% of circulating cases (26), but further work is needed to assess its feasibility for implementation across a wide range of settings. We also emphasize that classification of Absent-Vulnerable and Absent using this tool is only putative and does not certify or guarantee rabies-free status. However, we consider that this classification would be a strong indication of readiness, when a country or state is well positioned to compile evidence of freedom in line with OIE requirements (11).
We recommend applying this tool over a five-year window of surveillance data, as shorter periods tend to magnify transient patterns and longer windows are potentially less responsive to epidemiological transitions. Case data should be reported by state and municipality, as this also provides a simple criterion for assessing surveillance quality. In Mexico and Brazil, case locations were complete at state-level from 2011 onwards and therefore did not affect later classifications; however, classifications earlier in the study may have been affected by this missing information. Municipality information could also be useful as elimination is approached; in fact, we suggest that incorporating municipality information into inference approaches could generate a better understanding of incursion risks and more tailored management recommendations to prevent them. Future work testing the application of this tool in other countries and at other spatial scales would be valuable to better understand its utility.
Surveillance implications
With the progressive control and elimination of dog-mediated rabies from the region, other circulating virus variants have become increasingly apparent (37-39). Variant identification played a discriminatory role for our classifications of states in both Mexico and Brazil, where both dog- and wildlife-associated variants have co-occurred. Circulation of wildlife variants is not necessarily an obstacle to elimination of dog rabies variants, but wildlife variants have potential to spread in the dog population and pose public health risks. Strategies for maintaining rabies freedom and for judicious use of post-exposure prophylaxis therefore need to account for these potentially complex situations, which may affect the scaling back of dog vaccination and protocols for identifying and treating bite victims. We therefore recommend that as states approach elimination (Intermittent, Absent-Vulnerable or Absent classification), sequencing of all detected cases be undertaken (Figure 4). Since only very few cases are detected in these situations, this should not be cost prohibitive, though, baseline characterization of historically circulating variants is required.
The target for both Mexico and Brazil is now nationwide interruption of transmission and verification of disease freedom. For diseases that have been eradicated or regionally eliminated, intensified surveillance approaches have been employed to increase case detection (40-42). Such approaches are now urgently required for areas classified as Absent or Absent-Vulnerable, or with Intermittent detection, to resolve uncertainties regarding viral circulation, initiate early outbreak responses and verify freedom (Box 2). We suggest that surveillance guided by Integrated Bite Case Management (43), with epidemiological investigations triggered by bites from suspicious animals, should enable verification of rabies freedom and guide scaling back of mass dog vaccination (26). This approach should also help to identify transmission of other rabies variants and ensure appropriate treatment for exposed persons (44). Moreover, this approach should result in sample submissions, as well as observation or quarantining of biting animals (45) which are important surveillance indicators. Likewise, genomic surveillance is a valuable tool (46), that used in combination with epidemiological data, could be crucial in guiding the rabies endgame. Sequencing of viruses can resolve key questions about viral circulation, discriminating wildlife variants from dog variants (37), including the potential for host shifts (47), and providing insights about the persistent lineages in remaining foci (48), which would be useful now in Chiapas state, Mexico and in Guatemala. In a previously rabies-free area, sequencing could identify the source(s) of incursions (49), and confirm that the new virus lineage differed from those historically circulating (i.e. undetected endemic circulation) (30, 50). Adding functionality within SIRVERA for mapping georeferenced sequence data could facilitate rapid assessment of circulating viral lineages and potential incursion threats. Finally, contingency plans are urgently needed for states approaching elimination, and these should be regularly reviewed to ensure response capacity is maintained (Figure 4).