Progress towards one health approach for dog-mediated human rabies elimination in Bangladesh: on the way to zero by 30

doi:10.21203/rs.3.rs-3201454/v1

Download PDF

Article

Progress towards one health approach for dog-mediated human rabies elimination in Bangladesh: on the way to zero by 30

https://doi.org/10.21203/rs.3.rs-3201454/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this older preprint version

Read the latest preprint version →

Bangladesh is making progress toward achieving zero human dog-mediated rabies deaths by 2030, a global goal set in 2015. We conducted a comprehensive analysis between the years 2011 to 2023 to understand the effectiveness of rabies control programmes and predict human rabies cases in Bangladesh by 2030 using time-series forecasting models. The human-to-dog ratio in Bangladesh was found to be 85.83, with an average dog population density of 11.51 dogs/km². The mass dog vaccination (MDV) campaign has resulted in the vaccination of 81.46% of the estimated 3,030,662 dogs in the country. A decreasing trend from predicted and observed human rabies cases was identified, and it is projected that Bangladesh will have zero human rabies cases by 2030 if the current trend persists. The phylogenetic analysis shows that rabies viruses in Bangladesh belong to the Arctic-like-1 group. Bangladesh's efforts demonstrate that eliminating dog-mediated human rabies is achievable with a One Health approach.

Health sciences/Diseases/Infectious diseases/Viral infection

Health sciences/Medical research/Epidemiology

Rabies, a zoonotic viral disease primarily transmitted to humans through the bite of infected dogs, kills an estimated 59,000 people each year worldwide and has the highest case fatality rate of any infectious disease^1,2. More than 15 million human exposures to rabies occur annually, primarily in Asia and Sub-Saharan Africa, causing 3.7 million Disability-Adjusted Life Years (DALYs) and an enormous economic loss of USD 8.6 billion¹. Though the World Health Organization (WHO), the World Organisation for Animal Health (WOAH), and the Food and Agriculture Organization of the United Nations (FAO) have set a goal for the elimination of dog-mediated human rabies globally by 2030, there are few instances of extensive dog vaccination programmes that have successfully achieved this goal outside of Africa and Asia³.

Rabies has high public health consequences for Bangladesh which ranks third among rabies-endemic countries in the world in terms of human rabies deaths, about 96% of which are attributable to dogs^4,5. The Government of Bangladesh (GOB) launched the National Rabies Elimination Programme in 2011 and is taking necessary actions to achieve the goal of Zero by 30 through practical progress in rabies control⁶. GOB was instrumental in driving the programme forward by implementing strategies, including advocacy, communication, and social mobilization (ACSM), modern treatment for dog bites, mass dog vaccination (MDV), and dog population management⁷. This programmes reduced the number of deaths from human rabies by approximately 50 percent in recent years in Bangladesh⁸.

The early administration of human post-exposure prophylaxis (PEP) can prevent rabies-related deaths but treating people (a dead-end host) after being bitten has little effect on the incidence of rabies in the canine reservoir population, leaving additional members of the community at risk of contracting the infection⁹. For over a century, it has been known that MDV may successfully eliminate rabies from the reservoir canine population, preventing virus transmission to humans and eliminating dog-mediated rabies in many regions^9,10. In a One Health approach, modern rabies management emphasizes the necessity of attaining zoonotic disease prevention and control by considering human, animal, and environmental factors¹¹. MDV is an effective method to stop dog-to-dog or dog-to-human transmission of the rabies virus. If annual pulse vaccination campaigns achieve over 70% coverage, rabies incidence in dogs is likely to be significantly reduced, and regional elimination is possible if this coverage is sustained over several years^12,13. Because of the particular ecology of dogs in Bangladesh, where millions of dogs roam freely and are difficult to reach⁴, rabies is thought to be particularly challenging to eliminate, as evidenced by the complete paucity of examples of rabies elimination in any region of Bangladesh⁶.

It is crucial to evaluate the available data regularly at both the regional and global levels to adequately track the advancement toward the eventual goal of Zero by 30. The time series approach is commonly used to analyze a series of data points that are ordered in time and is used for dealing with data that may include trends and seasonality¹⁴. Thus, time series analysis may be appropriate for developing a predictive model for rabies in humans¹⁴ and a potent tool in the armory of attempts to prevent and control the disease because it can be used to predict critical epidemiological characteristics, set control targets, and inform policy¹⁵.Given the global goal of eliminating human deaths from dog-mediated rabies by 2030, models of rabies virus transmission have the potential to inform control efforts as countries continue to work toward elimination. While Bangladesh is marching towards the shared goal of Zero by 30, information on rabies trends and forecasting is crucial. We sought to investigate the trends and predict the human rabies cases as related to the use of MDV and human anti-rabies vaccine (ARV) in Bangladesh. Furthermore, we have compared the phylogenetic analyses of rabies virus sequences reported in Bangladesh and other South Asian countries extracted from GenBank.

MDV, ARV, and human rabies

We subjected 12 years (November 2011–April 2023) of MDV data maintained at Bangladesh's Directorate General of Health Services (DGHS). Bangladesh's MDV programme began in 2011 in the municipality of Cox's Bazar on the country's southeast coast. The campaign quickly swept throughout the country. When it comes to vaccinating free-roaming dogs, collaboration is vital. The One Health concept was implemented by bringing together experts from diverse sectors and levels of government. During this period, MDV was conducted at least once per year and was scaled up for at least one round in all 64 districts of the country. Thirty-five districts have already completed their second round of vaccination, while eight districts have completed their third round of vaccination. From 2011 onward, an upward trend of MDV was found, and the highest number of dogs was vaccinated in 2019 (n = 625,208) (Fig. 1, Supplementary Fig. 1). Under the MDV campaign, out of an estimated 3,030,662 dogs, an estimated 81.46% (95% CI: 80.54–82.37) have been vaccinated. The average dog population density in Bangladesh is 11.51 dog/km² (95% CI 10.09, 12.92), and the human-to-dog ratio is 85.83 (95% CI 74.63, 97.03) (Fig. 2). From January 2011 to April 2023 a total of 273,0829 doses of ARV were used to manage the animal bite cases at the National Rabies Prevention and Control Centre (NRPCC), the Bangladesh Institute of Tropical and Infectious Diseases (BITID), and the District Rabies Prevention and Control Centres (DRPCCs) of Bangladesh with an average of 18451.55 doses ARV used each month (Fig. 3). In total, 859 cases of human rabies were reported in Bangladesh between January 2011 and April 2023, with an average of 5.80 cases recorded monthly (0.0034 cases per 100,000 population) and average seventy cases recorded annually (0.04 cases per 100,000 population). The highest cases were reported in 2014 (106 cases, 6.8 cases per 10 million population). The districts in the middle of the country (the capital and nearby district) saw the highest number of cases (Fig. 4). We also found an increasing trend of ARV utilization and dog vaccination as well as a decreasing trend of human rabies cases in Bangladesh from 2011–2022 (Supplementary Fig. 1).

Two main clusters were revealed by the dendrogram of the rabies virus in Bangladesh (Fig. 5). The left main branch displays the first strong cluster. This cluster exposed strong to moderate positive correlations between three variables (MDV, ARV, and the year in which the vaccinations occurred). We have found strong positive correlations between years of MDV and ARV (Correlation coefficient: 0.60). Weak positive correlations have been found between ARV and MDV (Correlation coefficient: 0.32), whereas MDV and year of vaccination showed moderate positive correlations (Correlation coefficient: 0.41). The next main cluster is the right main branch, in which there were moderate negative correlations for human rabies death. Human rabies death and year of vaccination have a moderate negative correlation (Correlation coefficient: -0.56), whereas MDV and ARV and human rabies death were comparatively found to have a weaker negative correlation (Correlation coefficient: -0.20; -0.19), respectively.

Phylogenetic tree

Coordinated rabies control efforts in the area continue to be hampered by a need for more knowledge regarding the phylogenetic relationships between the rabies virus in Bangladesh and viruses in other countries. To explain its epidemiologic relationships, origin, and transmission dynamics, a phylogenetic analysis was performed to characterize the rabies virus circulating in Bangladesh and determine its relationship with viruses in neighboring countries. The phylogenetic tree revealed that the rabies viruses in Bangladesh are members of the Arctic/Arctic-like virus, also called Arctic-like-1. N gene sequences from rabies viruses in Bhutan and Bangladesh show a strong link, suggesting that they shared an ancestor and developed into distinct strains. The exact lineage circulated in other countries in this region, including India, Nepal, Pakistan, and Afghanistan, as shown by the tree samples clustered separately from sequences generated in other studies (Fig. 6).

Time-Series Analysis

The years 2006, 2007, and 2008 had the highest number of human rabies cases, with 167, 166, and 165 cases, respectively. On the other hand, the lowest occurrence of cases was observed in 2020, 2021, and 2022, with 26, 38, and 45 cases, respectively. Initially, there was a declining trend in cases from 2006 to 2013, followed by an increase in 2014, but the trend then resumed a downward trajectory. However, starting from 2015, the trend fluctuated, with periods of both improvement and decline (Fig. 7 (a)). In recent years, starting from 2021, there has been a slight increase in rabies cases, rising from 38 in 2021 to 45 in 2022. From January 1, 2023, to April 30, 2023, there were 17 confirmed human rabies cases reported in Bangladesh. When examining the monthly data, it was observed that rabies cases were most prevalent between November and February. The highest number of rabies cases occurred in December 2009 (n = 23), followed by December 2008 (n = 22), January 2009 (n = 20), November 2009 (n = 20), and December 2006 (n = 19), making them the top five months with the highest rabies incident rates.

The time series data representing rabies cases were analyzed by decomposing it into three components: trend-cycle (Tt), seasonal (St), and error/residual (Et). To determine the appropriate polynomials (p and P) and (q and Q) for the SARIMA model, the autocorrelation function (ACF) and partial autocorrelation function (PACF) correlograms were examined (Fig. 7 (b)). These correlograms assisted in specifying and identifying the suitable parameters for the SARIMA model.

Based on the results of significance tests and the analysis of residual correlograms using the ACF and PACF, the SARIMA (2, 2, 2) (1,0,0) [12] model was estimated. The correlograms showed that all lag values fell within the confidence interval, confirming the chosen model (Fig. 7 (c)).

By analyzing the time-series frequency and utilizing the observed number of rabies cases from January 2006 to April 2023, we observed a decreasing trend with slight seasonality in human rabies cases. The SARIMA model demonstrated strong predictive capabilities, as evidenced by the R2, RMSE, and MAE values of 94.22%, 2.72, and 2.03, respectively. The model identified a consistent downward trend between observed and predicted human rabies cases (Table 1). The model's projections suggest that Bangladesh is expected to experience no human rabies cases by 2030 if the current trend of decreasing rabies cases continues (Fig. 7 (d)).

Table 1

Prediction* and summary of SARIMA model using rabies cases from May 2023 to December 2030
SARIMA (2,2,2) (1,0,0) [12]
	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
2023					2	3	2	2	2	2	2	2
2024	2	2	2	2	2	2	2	2	2	2	2	2
2025	1	1	1	1	1	1	1	1	1	1	1	1
2026	1	1	1	1	1	1	1	1	1	1	1	1
2027	1	1	1	1	1	1	1	1	1	1	1	1
2028	1	1	1	0	0	0	0	0	0	0	0	0
2029	0	0	0	0	0	0	0	0	0	0	0	0
2030	0	0	0	0	0	0	0	0	0	0	0	0
RMSE: 2.72	MAE: 2.03	R2 (%): 94.22
*Rounded values

In the count time-series generalized linear model (TSGLM), the impact of each variable is represented by the incidence risk ratio (IRR). According to this model, it was observed that with every unit increase in MDV, the relative risk of rabies cases decreased by 1% [IRR = 0.99 (95% CI: 0.99–1.01)]. Similarly, for each unit increase in ARV, the relative risk of rabies cases in Bangladesh decreased by 1% [IRR = 0.99 (95% CI: 0.98–0.99)] (Table 2).

Table 2

Risk ratio of MDV and ARV to rabies cases in Bangladesh using time-series GLM methods
	IRR (95% CI)	P-value
MDV	0.99 (0.99–1.01)	0.504
ARV	0.99 (0.98–0.99)	< 0.001

This integrative rabies control programme using the One Health approach demonstrates the viability and sustainability of such a program in Bangladesh, providing a tangible example of achievement in eliminating dog-mediated human rabies deaths by 2030. Although rabies takes a terrible toll in Bangladesh, the inaccessibility of the vast population of free-roaming dogs has led to the current discourse that, while local rabies eradication within restricted communities is feasible, extending these approaches to a country level remains a nearly insurmountable practical and logistical challenge^11,16,17. As a result of such failures, the scientific community has been urged to refocus rabies research on initiatives that promote the field application and evaluation of rabies elimination strategies^9,18.

We employed time-series forecasting models using reported cases in the near past. Few, if any, detailed studies have been published on the forecasting of human rabies in Bangladesh. Infectious disease modeling is employed to comprehend the disease's dynamics better and assist with developing control strategies. Additionally, it might reveal how dynamics, consequently, intervention tactics may alter as management is put in place^15,19. Based on the models, we identified a declining trend with slight seasonality in human rabies cases from January 2006 to April 2023. The SARIMA model demonstrated excellent predictive abilities, with R2, RMSE, and MAE values of 94.22%, 2.72, and 2.03, respectively. When comparing the observed and predicted human rabies cases, the model suggested a consistent decrease in numbers. Based on the model's projections, if this downward trend in human rabies cases continues, Bangladesh could potentially eliminate dog-mediated human rabies cases by 2030. Model predictions and surveillance studies with similar socioeconomic settings showed that dog vaccination could save human lives and prevent considerable human rabies exposure over a certain period²⁰. Our previous study on rabies in Bangladesh identified a trend of declining human rabies cases that can be explained by the increased number of dogs being vaccinated against rabies⁶. Countries with 70% or higher vaccination coverage were classified as Phase III, and it was suggested that countries vaccinate 70% of the canine population for seven years to eliminate dog rabies^1,21,22. Lack of knowledge that dogs needed rabies vaccinations and ignorance of where to find the vaccine were frequently cited as vaccination barriers²³.

One of the challenges to increasing the amount of MDV in Bangladesh could be budget allocation, but there is a strong political commitment to eliminate dog-mediated human rabies⁸. To achieve the global goal of zero by 30, Bangladesh needs to increase dog vaccination substantially, and resources from national and international organizations would be an essential decider of achieving the goal on time. Although dogs were responsible for most of the exposures, cats also contributed significantly to the incidence of rabies in Bangladesh (12.11%), suggesting the inclusion of cats in the vaccination programme²⁴. It has been demonstrated that rabies prevention strategies, such as vaccination, are influenced by disease prevalence and that a simple model with intervention responses can accurately depict the host dynamics and periodicity of the disease²⁵. Many other factors can influence the occurrence of rabies cases, including literacy rate, GDP, awareness of rabies, access and availability of post-exposure vaccines, access to primary care, and primary contact related to sickness^6,26,27. However, reliable data on these parameters or proxy variables are not available. Thus, we decided to keep MDV and ARV in the model as the key determinant of the occurrence of human rabies cases. After MDV was introduced in Bangladesh in 2011, the pattern of human rabies cases altered quite quickly. The annual incidence of rabies in Bangladesh has consistently decreased as a result of the combined effects of a mass awareness campaign, PEP, and MDV⁶.

We observed that rabies cases were most prevalent between November and February. The highest number of rabies cases occurred in December 2009 (n = 23), followed by December 2008 (n = 22), January 2009 (n = 20), November 2009 (n = 20), and December 2006 (n = 19), making them the top five months with the highest rabies incident rates. Corresponding trends have been observed in previous research in Bangladesh, where the lowest instances have been identified in June and the greatest in January, with peaks in December and January and then again in April. However, there was no association observed in the study between monthly temperature or rainfall and the number of cases of animal bite²⁸. We found, however, that the majority of dog bite cases occur between the spring and summer months in developed countries²⁹. Such observations have been attributed to behavioral changes: greater interaction between pets and children during the warmer months, with less parental monitoring, increasing the chance of biting accidents³⁰. A study in the African region showed that the highest human rabies exposures were reported in spring (April to June), followed by winter (January to March), with autumn (October to December) having the lowest incidence of human rabies exposure³¹.

The phylogenetic tree showed that the rabies viruses in Bangladesh are part of the Arctic/Arctic-like virus, and the N gene sequences from the rabies viruses in Bhutan and Bangladesh exhibit a significant link, indicating that they shared an ancestor before evolving into different strains. AAL2 spread into central Bangladesh 32.3 years ago (95% HPD 18.4–50.6 years) around 1978 (95% HPD range 1958–1991), where glycoprotein has three potential N-glycosylation sites influencing viral pathogenesis³². Separate lineages were also discovered in other countries in this region, including Iran, Nepal, Pakistan, and Afghanistan, according to some studies^9,32. The rabies virus's diversity may have public health consequences in Bangladesh. Given the ease with which people can travel between countries, AAL2 most likely entered Bangladesh from India rather than Bhutan.

Our analysis indicates that the average dog population density in Bangladesh was approximately 12 dogs per km², with a human-to-dog ratio of approximately 86. A prior study in Bangladesh revealed 14 dogs per km², with a human-to-dog ratio of 120³³. The high ratio may be attributed to the abundance of edible waste on the streets, societal tolerance of stray dogs, and a lack of consistently implemented long-term birth control programmes³⁴. While Bangladesh has almost three times the benchmark number of dogs per km2 to become endemic for rabies, the threshold density for rabies persistence is only 4.5 dogs per km2^33,35. Bangladesh's dog population density is comparable to other Asian³⁵ and African countries³⁶.

From 2011 onward, we identified an increasing trend of human ARV utilization and MDV followed by a decreasing trend of human rabies cases in Bangladesh. After MDV was introduced in Bangladesh in 2011, the pattern of human rabies cases altered quite quickly^6,8. In recent years, the number of human rabies fatalities has decreased by approximately 50%, and annual cases have decreased at a rate of approximately 12 per year, while the number of vaccinated dogs has grown at a rate of 3200 dogs per year^6,8. The annual incidence of rabies in Bangladesh has consistently decreased due to the combined effects of a mass awareness campaign, PEP, and MDV. This accomplishment results from coordinated efforts by the Ministries of Health and Family Welfare, Fisheries and Livestock, Local Government, Rural Development & Cooperative, and Education. Together, they carried out rigorous MDV campaigns, offered PEP to animal bite victims, and increased awareness nationwide through ACSM. Working together across sectors is essential to achieving the Zero by 30 goals and is a shining example of what a One Health approach can accomplish¹⁶. All 64 of Bangladesh's districts now have at least one facility that offers PEP and wound care to people who have been bitten by animals⁶. Additionally, Bangladesh created more than 300 sophisticated Animal Bite Management Centres at the national, district, and sub-district (Upazilla) levels, in addition to preventing rabies at the source (the dogs). It is critical to guarantee that distant communities have access to medical treatment. The decision to visit one of the bite treatment centres is heavily influenced by travel time and accessible transportation options. They are now ensuring an ongoing supply of human rabies vaccine and immunoglobulin to deliver rabies post-exposure prophylaxis to over 400,000 bite sufferers each year¹⁶. More than 250,000 people have received this care from skilled nurses and doctors without paying anything out of pocket. The rise of MDV has confirmed the potential for successful rabies elimination strategies to result from a multi-sectoral, One Health strategy combining innovation, capacity-building, and broad implementation^6,8.

Our findings demonstrated that some simple known interventions could help address crucial public health issues and how a genuine One Health approach should be used to control zoonotic diseases like rabies. Additional factors, such as increased awareness of rabies and availability of PEP for animal bite victims, along with MDV, may contribute to changes in the detection of human rabies cases in Bangladesh⁶.

Our investigations had some limitations. Based on the cases that have been identified at public hospital facilities (NRPCC/DRPCCs), we determined the mortality rates for human rabies, which limited our scope. We are conscious of the possibility that some cases might have been overlooked, particularly those who sought care at private hospitals or from traditional healers and never went to a public hospital. However, we think that only a small number of rabies deaths have been missed by the government's primary data-collection facilities thanks to advancements in education and awareness. Due to the patients' reliance on their relatives for historical recollections, there may have been recall bias. However, we endeavored to lessen this bias by speaking with some of the patients' relatives on the phone. Even though there are issues like these, our research has provided insightful information on rabies in Bangladesh and can be applied to rabies control in countries with similar socioeconomic settings.

MDV programmes have shown positive outcomes, with an increase in vaccination and a decrease in human rabies cases in Bangladesh from 2011–2022. The trend in human rabies cases has fluctuated but could be eliminated if the decreasing trend continues. Increasing MDV and ARV led to a decline in the relative risk of human rabies cases. The rabies virus circulating in Bangladesh belongs to the Arctic/Arctic-like virus, and experts should continue coordinating vaccination efforts. Our findings demonstrate that successful MDV programmes operations are feasible in Bangladesh and can eliminate dog-mediated human rabies across broad geographic areas. The results of our research can be used to support national rabies control planning in Bangladesh and other nations with comparable socioeconomic environments, accelerating similar examples of success in achieving the 2030 target of eliminating dog-mediated human rabies worldwide.

Data Source

We obtained data from patient immunization record books maintained at the National Rabies Prevention and Control Centre (NRPCC) located in the Infectious Disease Hospital (IDH) within Dhaka, Bangladesh, from January 2011 through April 2023. As the primary referral center for individuals suffering from animal bites and rabies in Bangladesh, the NRPCC provides free vaccination and treatment to most animal bite cases that arise throughout the country⁶. Another centre similar to the IDH in the southern city of Chittagong is the Bangladesh Institute of Tropical and Infectious Diseases (BITID), the secondary referral centre for rabies after NRPCC. Besides NRPCC and BITID, 66 public District Rabies Prevention and Control Centres (DRPCCs) exist, with at least one centre in each of the 64 districts, which provide free ARV and treatment to individuals who have suffered animal bites. For data regarding MDV, we used the MDV database of the CDC, Bangladesh's Directorate General of Health Services (DGHS), from November 2011 through April 2023. The database includes data on all estimated and vaccinated dogs, human-to-dog ratios, and vaccination coverage in all 64 districts of Bangladesh.

Human Rabies Surveillance

The Directorate General of Health Services has a central core committee overseeing human rabies surveillance activities. This committee is led by the Director of Disease Control and Line Director, Communicable Disease Control (CDC), as well as additional personnel like the Deputy Director, Deputy Programme Manager, and Surveillance Medical Officer. The Upazila (sub-district) Rapid Response Team (URRT), comprised of medical professionals, health assistants, and other health personnel, has been investigating reports of patients who have had contact with animals (typically bites or scratches). The CDC and NRPCC's central rabies teams and the intermediate-level (Civil Surgeon) rabies teams have regularly received information and aggregate statistics from the local level.

MDV

The Communicable Disease Control Division (CDC) of DGHS coordinated the MDV programmes as part of the national rabies elimination programmes. They also supplied required dog rabies vaccines for MDV. Along with the DGHS, the MDV programmes in Bangladesh also received contributions in various forms from the Department of Livestock Services, the Local Government Division, the Education sectors, and non-governmental organizations (NGOs) or development organizations⁶. Vaccination teams were formed and trained through animal control staff training conducted by MDV experts prior to the MDV campaign. Vaccination team direction and programme monitoring were preplanned in a programme prior to conducting each vaccination campaign, called 'microplanning'. Each site will consist of 2–3 days of planning, training, and community mobilization, 3–4 days of vaccination, and one day of follow-up and evaluation. The goal was to vaccinate at least 70% of the dog population annually using a Capture Vaccinate Release (CVR) method¹. Each dog received a dose (1 ml) of the rabies vaccination (CaniShot RV-K -(CAVAC); CANVAC® R - Dyntec; Rabisin® - Merial) that was given either subcutaneously or intramuscularly, depending on the animal's posture and mode of restraint. To make it easier to identify the vaccination status on post-vaccination surveys, non-toxic paint was used to mark each dog on the top of its head. The paint was designed to last for several days. Prior to vaccinating dogs that could be clearly attributed to an owner, consent was acquired from the owner. Typically, a vaccination coverage evaluation is conducted on vaccination day 2 or 3, vaccination continued if necessary for another 1–2 days, and then a final coverage evaluation is completed.

At the time of vaccination, the following information was recorded for each dog: vaccination team ID, date, and time vaccine was given, GPS coordinates, sex and approximate age of dogs, ownership status (if known), neuter status, confinement, previous vaccination history, and health status/body condition status. An electronic data capture device (i.e., GPS data logger or mobile phone application) was also used to assist with data collection and final evaluation. In November 2011, MDV was first piloted in Bangladesh, covering a small municipality. Following the pilot's success, more comprehensive MDV campaigns were implemented nationwide⁶.

Since 2011, CVR is currently practiced as the primary mass vaccination strategy in Bangladesh. CVR teams are typically comprised of five or six members who travel by vehicle: one vaccinator, three dog catchers who use nets, one driver, and one rapporteur. Dogs that could be held in the owner's or team's hands were manually restrained for vaccination. Nets were used to catch dogs that could not be restrained manually.

It is expected that two vaccination teams can cover an area containing roughly 200 dogs in 3 days. Previous studies estimated a dog density of 52 dogs/km² in Dhaka³⁷ so that each urban zone would be about 3.8 km² (1.5 mi²). Previous studies also estimate a human: dog ratio of 120:1 and a density of 14 dogs/km² in rural areas³³, so rural zones will be defined based on a population of 24,000 people per zone, or about 14 km² (5.4 mi²).

Post-vaccination dog survey was conducted to estimate the coverage of free-roaming dogs in the community, and the dog population was estimated using the 'Capture-mark-recapture' method³⁷.

The estimation of the dog population was done using the Lincoln-Petersen's formula with Chapman’s correction: $N=[\frac{({n}_{1}+1)({n}_{2}+1)}{(m+1)}$] – 1 ……………… (1)

Where, n denotes the calculation of the entire population's size, n₁ signifies the combined sum of dogs sighted on the 1st day, n₂ signifies the combined sum of dogs sighted on the 2nd day, and m signifies the number of marked dogs seen on day 1 that were also sighted on day 2.

An almost unbiased variance of N was estimated using Seber's formula³⁸.

$$var\left(N\right)=\left[\frac{({n}_{1}+1)({n}_{2}+1)({n}_{1}-m)({n}_{2}-m)}{{(m+1)}^{2}(m+2)}\right]$$

………………

and the 95% confidence interval for N is estimated as:

$$N\pm 1.965 \sqrt{var\left(N\right)}$$

………………

Dog vaccination coverage for free-roaming dogs was estimated from the proportion of dogs counted after the vaccination campaign that vaccinators marked.

Statistical Analysis

We also performed Spearman rank correlation coefficients between key rabies patient characteristics and ARV variables to identify the association between the variables. Hierarchical clustering using Pearson correlation coefficients was used to visualize associations between variables and possible clusters using pairwise comparisons to adjust correlations to measure dissimilarity and the distribution function in R software.

Time-Series Analysis

Earlier observations potentially influence any variable measured across time, an essential aspect of time-series analysis (autocorrelation). Time-series models use previous observations as the foundation for forecasting future behavior to take advantage of these linkages. This is the significant difference between time-series analysis and conventional statistical tests for assessing change, such as regression analysis, which relies on independent variable variation to explain outcome³⁹ changes 38. A time-series forecasting model called the seasonal autoregressive integrated moving average (SARIMA) model was employed⁴⁰ 39 to forecast rabies cases based on monthly data. This model combines various components, namely the trend cycle, seasonality, and error, and can be represented by the following equation:

Zt = Tt + St + Et

Here, Zt refers to the observed data, Tt represents the trend-cycle component, St represents the seasonal component, and Et accounts for the portion of data not accounted for by the model in a given period, also known as the error or random residual.

As the primary outcome variable for rabies cases is influenced by past reported cases with some seasonality, we chose the SARIMA model (time-series events). The SARIMA model is a data-focused, exploratory technique that enables the user to construct an appropriate model based on the data's actual structure^40,41. This model attempts to extract regional trends while filtering out high-frequency noise from the data and assumes a linear correlation between the time series values⁴². By updating the model to forecast the system's future state based on current occurrences, SARIMA models have the advantage of adapting to dynamically oriented systems that change over time. The R package "forecast" was used to run the SARIMA model for this study⁴³. We predicted trends for the upcoming 7 years (up to 2030) using SARIMA time series models with the data on rabies cases and showed them in the figure.

The Box and Jenkins approach was utilized in formulating the seasonal autoregressive integrated moving average (SARIMA) model, which is intended for analyzing stationary time series data^40,44,45. The SARIMA model is characterized by the notation SARIMA (p,d,q)(P, D, Q), where p and P represent the order of the non-seasonal autoregressive polynomial and seasonal autoregressive polynomial, respectively. Similarly, q and Q denote the order of the non-seasonal moving average polynomial and seasonal moving average polynomial, respectively. The autoregressive and moving average orders were selected by examining the autocorrelation function (ACF) and partial autocorrelation function (PACF). Furthermore, d and D represent the orders of differentiation applied to the time series to achieve stationarity. A SARIMA(p,d,q)(P, D, Q) process refers to an autoregressive moving average (SARMA) model that has been differenced d and D times to obtain stationarity.

Numerous researchers have recently used time series to study the trend of rabies (including additive models⁴⁶, auto-regressive time series models⁴⁷, and wavelet time series models²⁵); however, most of them use the ARIMA model with one-time series because it is a helpful method for analyzing time series with one-time series in systems^14,48. Because of this, we used the SARIMA model to capture the seasonality of the outcome variable. SARIMA models accept a direct relationship between the time-series values and seek to leverage these straight circumstances in perceptions to extract nearby designs while reducing high-frequency turbulence. However, as SARIMA only considers one variable, it could not provide light on the connections between system variables.

We employed count time series following generalized linear models (TSGLM) to examine the relationship between our study's outcome variable, which consists of count data, and other variables. This class of models offers great flexibility and can effectively capture serial correlation while maintaining simplicity. The past observations, potential effects of covariates, and previous values are linked to the conditional mean of the observed process. Our research particularly concentrated on models that used the logarithmic link function as the conditional distribution conformed to a Poisson regression model.

In this study, Mass Dog Vaccination (MDV) and Anti Rabies Vaccination (ARV) were considered explanatory variables in the TSGLM model. The 'tsglm’ packages from R software were used for TSGLM analysis.

Empirical evaluation

The relevance of the predictions empirically assesses the SARIMA model; we analyzed and compared the performance of this time series model using some of the generally used measures, including coefficient of determination (R2), root mean square error (RMSE), and mean absolute error (MAE).

Cartographic display

The cartographic boundary files, which are used to create maps, were downloaded from the DIVA-GIS⁴⁹. All cartographic displays were performed in ArcGIS version 10.8.1⁵⁰. Choropleth maps were used to display the district-level geographic distribution of rabies cases, human-to-dog ratios, number of vaccinated dogs, and percentage of vaccination coverage using Jenk’s optimization classification scheme.

Phylogenetic Analysis

We used the complete genome sequences of the N gene of the arctic-like rabies virus from Bangladesh (GenBank accession number AB699214.1) and blasted in National Centre for Biotechnology Information (NCBI) to generate a phylogenetic tree. Then we downloaded 68 complete genome sequences of the N gene from Bangladesh, India, Pakistan, Nepal, Afghanistan, and Bhutan. The sequences were aligned by the MUSCLE method in MEGA 11 software. The N sequences' phylogenetic relationship was determined using the neighbor-joining method in MEGA 11. The maximum likelihood phylogenetic trees of the N segment were generated using the Hasegawa-Kishino‐YanoJukes model with a bootstrap value of 1000^51,52. Finally, the phylogenetic tree was visualized in FigTree version v1.4.3. The viruses from Bangladesh were labeled red.

Data availability

Data supporting the findings of this work are available within the paper and its Supplementary Information files.

Acknowledgements

The National Rabies Elimination Programme was funded by the Communicable Disease Control (CDC) Division of the Directorate General of Health Services (DGHS) of Bangladesh. We thank all the staff of DGHS, Department of Livestock Services (DLS) and Local Government, volunteers, and NGOs who have supported and contributed directly or indirectly to the success of the campaign. Use of trade names, product names, or commercial sources is for identification only and does not imply endorsement by the Government of Bangladesh.

Author Contributions

Conceptualization, S.G., M.N.H., and N.H.; Methodology, S.G., M.N.H., N.D., D.J., and N.H.; Data Analysis, M.N.H., N.D., D.J., M.J.U., and S.G.; Field Investigation, M.K.I., M.M.R., N.M., M.K., M.F.R., S.K., S.M.U., M.R.A.S., M.S.R., and A. A. J.; Data Processing, M.N.H., N.D., S.G. and H.S.M.; Writing – Original Draft, S.G., M.N.H., N.H. and S.C.; Supervision, T.S.S., M.N.I., S.M.G.K., B.A., and U.R.S.

Additional Information

Competing interests

The author(s) declare no competing interests.

World Health Organization & WHO. WHO Expert Consultation on rabies. Third report. World Health Organization technical report series, 1012 (2018). 2018.
Hampson K, Coudeville L, Lembo T, et al. Estimating the global burden of endemic canine rabies. PLoS neglected tropical diseases 2015; 9(4): e0003709.
Minghui R, Stone M, Semedo MH, Nel L. New global strategic plan to eliminate dog-mediated rabies by 2030. The Lancet Global Health 2018; 6(8): e828-e9.
Hossain M, Ahmed K, Bulbul T, et al. Human rabies in rural Bangladesh. Epidemiology & Infection 2012; 140(11): 1964-71.
Rana MS, Siddiqi UR, Ghosh S, et al. Epidemiological study of human rabies cases in Bangladesh through verbal autopsy. Heliyon 2020; 6(11): e05521.
Ghosh S, Rana MS, Islam MK, et al. Trends and clinico-epidemiological features of human rabies cases in Bangladesh 2006–2018. Scientific reports 2020; 10(1): 2410.
Ghosh S, Chowdhury S, Haider N, et al. Awareness of rabies and response to dog bites in a Bangladesh community. Veterinary medicine and science 2016; 2(3): 161-9.
World Health Organization. The Rabies Elimination Program of Bangladesh. Neglected tropical diseases. Geneva: WHO 2017, http://www.who.int/neglected_diseases/news/Bangladesh-rabies-elimination-program/en/, 2017.
Gibson A, Yale G, Corfmat J, et al. Elimination of human rabies in Goa, India through an integrated One Health approach. Nature Communications 2022; 13(1): 2788.
Kurosawa A, Tojinbara K, Kadowaki H, Hampson K, Yamada A, Makita K. The rise and fall of rabies in Japan: A quantitative history of rabies epidemics in Osaka Prefecture, 1914–1933. PLoS neglected tropical diseases 2017; 11(3): e0005435.
Rupprecht CE, Freuling CM, Mani RS, Palacios C, Sabeta CT, Ward M. A history of rabies—The foundation for global canine rabies elimination. Rabies: Elsevier; 2020: 1-42.
World Health Organization. Rabies vaccines: WHO position paper. Weekly epidemiological record; (93) 201–220, http://www.who.int/wer (2018). 2018.
Hampson K, Dushoff J, Cleaveland S, et al. Transmission dynamics and prospects for the elimination of canine rabies. PLoS biology 2009; 7(3): e1000053.
Paso A, Ngamjarus C. Forecasting rabies in dogs in Thailand: Time series analysis. Naresuan University Journal: Science and Technology (NUJST) 2020; 28(4): 64-74.
Heesterbeek H, Anderson RM, Andreasen V, et al. Modeling infectious disease dynamics in the complex landscape of global health. Science 2015; 347(6227): aaa4339.
Organization. WH. Mass Dog Vaccination- using the One Health approach to prevent rabies. https://www.who.int/southeastasia/news/feature-stories/detail/mass-dog-vaccination-using-the-one-health-approach-to-prevent-rabies#:~:text=Bangladesh's%20mass%20dog%20vaccination%20campaign,spread%20all%20over%20the%20country., 2022.
Bonwitt J, Bonaparte S, Blanton J, et al. Oral bait preferences and feasibility of oral rabies vaccination in Bangladeshi dogs. Vaccine 2020; 38(32): 5021-6.
Cleaveland S, Lankester F, Townsend S, Lembo T, Hampson K. Rabies control and elimination: a test case for One Health. Veterinary Record 2014; 175(8): 188-93.
Klepac P, Metcalf CJE, McLean AR, Hampson K. Towards the endgame and beyond: complexities and challenges for the elimination of infectious diseases. The Royal Society; 2013. p. 20120137.
Kunkel A, Jeon S, Joseph HC, et al. The urgency of resuming disrupted dog rabies vaccination campaigns: a modeling and cost-effectiveness analysis. Scientific reports 2021; 11(1): 12476.
Wallace RM, Undurraga EA, Blanton JD, Cleaton J, Franka R. Elimination of dog-mediated human rabies deaths by 2030: needs assessment and alternatives for progress based on dog vaccination. Frontiers in veterinary science 2017; 4: 9.
Cleaveland S, Kaare M, Tiringa P, Mlengeya T, Barrat J. A dog rabies vaccination campaign in rural Africa: impact on the incidence of dog rabies and human dog-bite injuries. Vaccine 2003; 21(17-18): 1965-73.
Yoak AJ, Haile A, O’Quin J, et al. Barriers and opportunities for canine rabies vaccination campaigns in Addis Ababa, Ethiopia. Preventive veterinary medicine 2021; 187: 105256.
Rana MS, Siddiqi UR, Ghosh S, et al. Epidemiological study of human rabies cases in Bangladesh through verbal autopsy. Heliyon 2020; 6(11).
Hampson K, Dushoff J, Bingham J, Brückner G, Ali Y, Dobson A. Synchronous cycles of domestic dog rabies in sub-Saharan Africa and the impact of control efforts. Proceedings of the National Academy of Sciences 2007; 104(18): 7717-22.
Subedi D, Chandran D, Subedi S, Acharya KP. Ecological and Socioeconomic Factors in the Occurrence of Rabies: A Forgotten Scenario. Infectious Disease Reports 2022; 14(6): 979-86.
Penjor K, Tenzin T, Jamtsho RK. Determinants of health seeking behavior of animal bite victims in rabies endemic South Bhutan: a community-based contact-tracing survey. BMC public health 2019; 19: 1-11.
Hossain M, Bulbul T, Ahmed K, et al. Five-year (January 2004–December 2008) surveillance on animal bite and rabies vaccine utilization in the Infectious Disease Hospital, Dhaka, Bangladesh. Vaccine 2011; 29(5): 1036-40.
Daniels DM, Ritzi RB, O’Neil J. Analysis of nonfatal dog bites in children. Journal of Trauma and Acute Care Surgery 2009; 66(3): S17-S22.
Tenzin, Dhand NK, Gyeltshen T, et al. Dog bites in humans and estimating human rabies mortality in rabies endemic areas of Bhutan. PLoS neglected tropical diseases 2011; 5(11): e1391.
Gebru G, Romha G, Asefa A, Hadush H, Biedemariam M. Risk factors and spatio-temporal patterns of human rabies exposure in Northwestern Tigray, Ethiopia. Annals of global health 2019; 85(1).
Jamil KM, Ahmed K, Hossain M, et al. Arctic-like rabies virus, Bangladesh. Emerging Infectious Diseases 2012; 18(12): 2021.
Hossain M, Ahmed K, Marma ASP, et al. A survey of the dog population in rural Bangladesh. Preventive veterinary medicine 2013; 111(1-2): 134-8.
Gill GS, Singh BB, Dhand NK, Aulakh RS, Ward MP, Brookes VJ. Stray Dogs and Public Health: Population Estimation in Punjab, India. Veterinary Sciences 2022; 9(2): 75.
Knobel DL, Cleaveland S, Coleman PG, et al. Re-evaluating the burden of rabies in Africa and Asia. Bulletin of the World health Organization 2005; 83: 360-8.
Kitala P, McDermott J, Kyule M, Gathuma J, Perry B, Wandeler A. Dog ecology and demography information to support the planning of rabies control in Machakos District, Kenya. Acta tropica 2001; 78(3): 217-30.
Tenzin T, Ahmed R, Debnath NC, Ahmed G, Yamage M. Free-roaming dog population estimation and status of the dog population management and rabies control program in Dhaka City, Bangladesh. PLoS neglected tropical diseases 2015; 9(5): e0003784.
Seber G. The effects of trap response on tag recapture estimates. Biometrics 1970: 13-22.
Linden A, Adams JL, Roberts N. Evaluating disease management program effectiveness: an introduction to time-series analysis. Disease Management 2003; 6(4): 243-55.
Dos Santos AJF, Ferreira JM, Baptista F, et al. Statistical analysis between 2006 and 2019 and forecast of rabies in cattle from 2020 to 2022 in Tocantins State (Brazil), by using the R Studio software. Epidemiology & Infection 2022; 150.
Saxena UR, Sharma P, Gupta G. Comprehensive study of machine learning algorithms for stock market prediction during COVID-19. Journal of Computers, Mechanical and Management 2022; 1(2): 14-21.
Adhikari R, Agrawal RK. An introductory study on time series modeling and forecasting. arXiv preprint arXiv:13026613 2013.
Hyndman R, Athanasopoulos G, Bergmeir C, et al. Forecasting functions for time series and linear models. R package version 2015; 6.
Quddus MA. Time series count data models: an empirical application to traffic accidents. Accident analysis & prevention 2008; 40(5): 1732-41.
Sato RC. Disease management with ARIMA model in time series. Einstein (Sao Paulo) 2013; 11: 128-31.
Olarinmoye AO, Ojo JF, Fasunla AJ, et al. Time series analysis and mortality model of dog bite victims presented for treatment at a referral clinic for rabies exposure in Monrovia, Liberia, 2010–2013. Spatial and spatio-temporal epidemiology 2017; 22: 1-13.
Lopes E, Sáfadi T, Da Rocha CMBM, Cardoso DL. Analysis of time series of cattle rabies cases in Minas Gerais, Brazil, 2006–2013. Tropical animal health and production 2015; 47: 663-70.
Bedi R, Verma N, Gautam K, Agiwal V. Forecasting the Anti-Rabies Vaccine Demand at Jawaharlal Medical College and Hospital, Ajmer, Rajasthan: A Comparative Analysis based on Time Series Model. Indian Journal of Community Health 2021; 33(3).
DIVA-GIS. Download data by country. https://www.diva-gis.org/gdata. Accessed 25 Jan 2023. 2023.
Introducing ArcGIS Enterprise 10.8.1. https://www.esri.com/arcgis-blog/products/arcgis-enterprise/announcements/arcgis-enterprise-10-8-1/. Accessed 25 January 2023.
Troupin C, Dacheux L, Tanguy M, et al. Large-scale phylogenomic analysis reveals the complex evolutionary history of rabies virus in multiple carnivore hosts. PLoS pathogens 2016; 12(12): e1006041.
Shimodaira H, Terada Y. Selective inference for testing trees and edges in phylogenetics. Frontiers in ecology and evolution 2019; 7: 174.

There is NO Competing Interest.

ARVMDVHuRabiesMonths20112023.xls
Dataset 1: ARV_MDV_Human_rabies
MDVCoverageinbangladeshUpdatetoMay2023.xlsx
Dataset 2: MDV_Bangladesh
Rabiescase20192023.xlsx
Dataset 3: Human_rabies cases_2019-2023
67centermasterfile2019202018.01.21.xlsx
Dataset 4
SupplementaryFile7.7.23.docx

Download PDF

Version 1

posted

You are reading this older preprint version

Read the latest preprint version →

Progress towards one health approach for dog-mediated human rabies elimination in Bangladesh: on the way to zero by 30

Status:

Version 1

Abstract

Figures

Introduction

Results

MDV, ARV, and human rabies

Phylogenetic tree

Time-Series Analysis

Discussion

Methods

Data Source

Human Rabies Surveillance

MDV

Statistical Analysis

Time-Series Analysis

Empirical evaluation

Cartographic display

Phylogenetic Analysis

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1