Page 44 - Read Online

Swedberg et al. One Health Implement Res 2023;3:77-96 https://dx.doi.org/10.20517/ohir.2023.02 Page 83

[24]
testing. Animal cases were confirmed either at RADDL or RITM with DFA, as recommended by WHO .
All samples collected during the study that tested positive by LFD were also confirmed positive
subsequently by DFA, but not all samples that were tested by DFA were also tested by LFD.

Patient risk categories were not updated from animal investigations, which could not be consistently
collected due to COVID-19 pandemic restrictions. Thus, for this study, patients were classified as either
low-risk, unknown-risk, or high-risk for rabies exposure based on the patient risk assessment from their
first visit to the ABTC, apart from laboratory-confirmed biting animals, for whom risk categories were
updated retrospectively. The risk of exposure categories used in this study was based on WHO animal case
definitions , where at least one criterion was met:
[1]

Low-risk: (WHO definition “Not a case”) Biting animal had no clinical signs of rabies and was healthy and
alive 14 days after the bite/exposure event or tested negative for rabies (if euthanized/killed).

Unknown-risk: (WHO definition “suspected” or “probable”) Biting animal not identified or found;
therefore, the history of the animal was unknown (e.g., vaccination/health status, contact with suspected,
probable, or confirmed rabid animal, health status, etc.).

High-risk: (WHO definition “suspected”, “probable” or “confirmed” animal case) Biting animal showed
clinical signs of rabies (e.g., aggressive/erratic behavior, hypersalivation, paralysis, tremors, abnormal
vocalization, loss of appetite); had a history of contact with suspect/confirmed rabid animal; and died within
14 days of exposure event; or tested positive for rabies.

Data analysis
Decision tree model
We used a decision tree framework to probabilistically describe the steps by which rabies infection in dogs
leads to human exposures and deaths, and associated costs. This type of framework has been used before to
estimate the burden of rabies [15,16] . Here, we extended the framework using IBCM data and further estimated
current surveillance performance and cost-effectiveness of prevention measures.

To simplify our analysis, we made several assumptions. We assumed that all bite patients who reported to
an ABTC received PEP (and that PEP was 100% effective in preventing rabies), considering that shortages
and vaccine refusal are rare in Oriental Mindoro. Additionally, we assumed that reported human rabies
deaths were recorded correctly with high probability (P obs|death ). Our estimates and 95% prediction intervals
(PrI) were based on 1,000 probabilistic draws of parameters described in Table 1, following the decision tree
framework.

IBCM risk assessment classifications were used to assign patients as either bitten by healthy dogs (low-risk)
or rabid dogs (high-risk), with uncertainty based on the observed range in IBCM risk assessments (lower
limits included only high-risk, while upper limits included high-risk plus unknown-risk). We used the
proportions of high-risk bites from both incomplete IBCM data and complete data from one ABTC to
extrapolate to the province. These estimates were compared to each other and resulting estimates from the
decision tree model.

Total exposures were calculated as the sum of high-risk exposures, assigned prospectively, that sought PEP
(from IBCM data) and estimated exposures that did not seek PEP extrapolated from recorded human rabies
deaths. Similarly, numbers of rabid dogs were estimated from total exposures and the average number of