Introduction

African swine fever (ASF) is a hemorrhagic disease of swine caused by the ASF virus (ASFV), a double-stranded DNA arbovirus of the family Asfarviridae1,2. The ASFV only infects swine, including domestic pigs, wild boar, warthogs, bushpigs, and other wild swine species. ASFV is not infectious to humans and does not pose a food safety threat, however, it causes high mortality in affected domestic pigs (nearly 100% with highly virulent strains). Because of its significant impacts to swine populations and severe economic consequences, ASF has been designated a reportable disease to the World Organisation for Animal Health (WOAH)3,4. Since 2007, ASF has been rapidly spreading globally and is currently present in Africa, Europe, Asia, South East Asia, and the Dominican Republic and Haiti1,5. ASF is difficult to control due to the lack of treatment or an approved and widely available vaccine, and consequently has caused severe disruption to the global swine industry in recent years3,5.

In 2021, ASF was detected in the Dominican Republic (DR) and Haiti6,7,8. Together, these countries make up the island of Hispaniola. Both countries were previously affected by ASF in 1978, which they dealt with by eradicating all susceptible swine on the island9. In response to the 2021 reintroduction, the DR veterinary authority, with international support, employed a series of emergency strategies focused on containing and eradicating the disease6. These control measures have included the development of laboratory capacity for ASF testing, surveillance, depopulation of infected herds, and compensation for culled animals. The ongoing outbreaks have had a significant impact on the DR’s swine industry, which is composed of over 13,000 producers that are mainly informal and backyard farms10. It has been estimated that production decreased by 21% in 2022 because of ASF compared to the previous year, and that approximately $28.7 million USD was spent on indemnification to affected producers for government-mandated depopulation from June 2021 to September 202311,12.

A recent evaluation of passive surveillance reports from the DR from November 2022-June 2023 suggested that, two years after its reintroduction, the disease is generally widespread throughout the country13. Evaluation of the reproduction ratio, a metric that describes outbreak growth or decline14,15, indicated that the ASF epidemic may be stabilizing in the population. However, this analysis was limited to only considering passively reported cases and does not provide information on the most recent months of the ASF outbreaks in the DR. Here, additional data from both passive reports and a recently initiated active surveillance program have been analyzed with the objective of describing the current status of the ASF epidemic in the DR, including estimating the cumulative incidence at country, region, and province level, and estimating the reproductive ratio of the outbreaks from late 2022 to early 2024. This work provides important information about the situation of the disease in the country for DR decision-makers in both the public and private sectors. These results will contribute to the design of prevention strategies to reduce ASF risk in the DR.

Methods

Data sources

Data on ASF surveillance from January 2023 to March 2024 was provided by the Ministerio de Agricultura Dirección General de Ganadería (DIGEGA) with support from the United States Department of Agriculture Animal and Plant Health Inspection Service (USDA-APHIS). The data received included the date that samples were received by the Ministerio de Agricultura Laboratorio Veterinario Central (LAVECEN), the name of the producer and/or farm name, the region and province of the farm, the type of surveillance (active or passive), and the result of real-time PCR (rt-PCR) diagnostic testing (positive or negative). These data represent official ASF testing in the Dominican Republic from January 2023 to March 2024. The swine population for each province in the DR was obtained from the 2022 national swine census conducted by the Ministerio de Agricultura and Banco Agricola with coordination and support from Organismo Internacional Regional de Sanidad Agropecuaria (OIRSA)10. This information is provided in Supplementary Table 1.

Surveillance reports came from both passive and active surveillance in the DR. Passive surveillance was from farms that reported suspicion of ASF to the DR veterinary authority based on the presence of clinical signs and were subsequently investigated following DIGEGA’s routine protocols for sampling (collection of whole blood, serum, and tissue samples and submission to LAVECEN for rt-PCR and enzyme-linked immunosorbent assay [ELISA]-antibody testing) and verification of outbreaks.

Active reports represent sampled farms without any specific clinical signs or suspicion for ASF as a part of ongoing government-initiated surveillance. Active surveillance reports were from two different sources: (1) mandated regular sampling from large commercial farms (see definition below), and (2) from an ongoing active surveillance program in backyard farms with less than 25 pigs. The former source of surveillance was instituted by DIGEGA in March 2023. It requires commercial farms with 25 or more pigs and with the highest levels of biosecurity in the country to submit approved samples (whole blood, serum, spleen, lymph nodes, or kidney) to LAVECEN every 21 days, uninterruptedly, for rt-PCR testing following LAVECEN’s standard laboratory protocols16. The second source of active reports, the ongoing active surveillance program, was designed and is being conducted by DIGEGA and USDA-APHIS. This program is intended to conduct surveillance in provinces where the number of samples received through passive reporting does not provide enough information to know the current status of disease in those populations. The sample size for each province and farm was estimated with a two-stage design using Cannon’s formula. Briefly, Cannon’s formula can be used to estimate the number of premises (farms) and the number of animals within premises to sample by applying the formula consecutively. The user assumes a confidence level, a prevalence threshold of disease, i.e., the minimum detectable prevalence, the number of premises/animals in the population, and the sensitivity and specificity of the diagnostic test17,18. Here, USDA-APHIS and DIGEGA veterinary services assumed a farm-level design prevalence of 2% and a within-farm design prevalence of 10% with 95% confidence. All backyard farms with at least one pig were eligible for sampling. Farms were randomly selected from information collected via the 2022 census. However, at the time of field sampling, many new producers were found and sampled by the field team. Whole blood samples were collected from selected farms and submitted to LAVECEN for ASF testing via rt-PCR. The active surveillance program began in September 2023, and by the time this manuscript was written in June 2024, surveillance was still ongoing. In the data that were provided for this analysis, the active surveillance source (mandated regular sampling or from the ongoing active surveillance program) was not specified and therefore could not be analyzed here.

Descriptive analysis

Descriptive analysis was performed in Microsoft Excel, version 2016, and R version 4.3.1. Counts and proportions were calculated for the number of reports, including temporally by month and week (aggregated on a Monday-to-Monday basis), surveillance type (active or passive), region, province, and result. Choropleth maps depicting counts or proportions for each province were produced using ArcGIS Pro version 3.2.219,20. Map breaks were selected using the natural breaks (Jenks) setting. The positive rate for each week was estimated as the number of positive reports in a given Monday-to-Monday week divided by the total number of reports in the same time period.

The cumulative incidence was estimated overall and for each region and province. The cumulative incidence is defined as the proportion of non-diseased individuals at the beginning of a time period that become diseased21. Here, this was calculated as the number of newly reported positive farms from January 2023-March 2024 divided by the total number of farms using the 2022 census estimates. There were five producers that had more than one positive report during the study period, and only their first occurrence of ASF was used to estimate the cumulative incidence. There were also 15 positive cases that were missing a producer name and were removed for the purpose of this analysis. In total, this left 206 new outbreaks (n = 25 removed), which were used to estimate the cumulative incidence for the country, by administrative region, and by province.

Because it was known that some farms would be sampled repeatedly during the study period (because of the requirement for samples every 21 days), the number of unique farms and the number of reports by farm were estimated. However, when samples were received by LAVECEN, farms were not consistently recorded using the same farm name or ID, and geographic coordinates were not available. Thus, without a clear way to identify repeating farms in the dataset, further analysis of the farm names was not performed and the estimated number of unique farms for each province are only approximations.

Estimation of reproductive ratio

The farm-level reproduction ratio (R0) was estimated for the country. R0represents the average number of secondary infections caused by one infected unit throughout the duration of the infectiousness period and can indicate whether an epidemic is worsening, improving, or stabilizing22. R0 > 1 indicates that each infectious unit causes more than one new infection and the outbreak is growing. R0 < 1 indicates that each infectious unit causes less than one new infection, and the outbreak is decreasing. R0= 1 indicates an infectious unit causes only one new case, and the outbreak size does not increase or decrease which can lead to endemicity14. The entire DR was considered the study area, and an infection was considered as an outbreak farm. R0 was estimated using all outbreaks in the current data (n= 231), and an additional 96 passively reported outbreaks from November 2022-June 2023. Briefly, these data were previously provided by USDA and DIGEGA and analyzed in a prior publication13. Together, these combined data allowed for estimation of R0 during the period from November 2022-March 2024 using 327 outbreaks.

R0 was estimated using the doubling time method, whereby R0is calculated for each time period in which the number of cases doubled15,23,24. However, R0 cannot be calculated for the end of the time period because the cases will not have doubled yet. Using this method, R0 was estimated as Eq. (1):

$$\:1\:+\:\left(\:\right(\frac{D}{td}\:*ln\left(2\right))$$
(1)

Where td=the time to double the number of cases and D = the duration of the infectious period of the farm (days). R0 was estimated for each doubling interval during the study period by assuming td=0 at the beginning of the time period (November 2022). Because it is difficult to estimate the duration of infection for an individual farm (as compared to an individual pig), a stochastic approach was used to account for natural variability and uncertainty in model parameters. The duration of infection was considered as the onset of infection in at least one individual pig on a farm until the depopulation of the premise. Three components were estimated: the latent (non-infectious) period of the first infected individual on the farm (Dlatent), the time for a farmer to detect and report ASF after the initial introduction (Ddetect), and the time for the veterinary authority to depopulate the farm after being notified (Ddepop). Then, D was estimated as Eq. (2):

$$\:D\:=\:Ddetect\:+\:Ddepop\:-\:Dlatent$$
(2)

Dlatent was parameterized as Gamma(19.9, 0.3919598; mean 7.8 days) based on previous estimations of transmission parameters using herd data25,26. Ddetect was parameterized as Pert(9,25,32 days), representing the minimum, most likely, and maximum time to detect ASF after introduction in a pen of 30 pigs with a total herd size of 1,200 from previous heterogenous transmission models27. Finally, Ddepop was estimated as Pert(0.5,1,3 days) using expert opinion from veterinary officials in the DR with extensive, first-hand experience in responding to the outbreaks28. The stochastic model was run in R version 4.3.1 using the mcd2 package and 10,000 Markov chain Monte Carlo runs29.

Because the doubling time method assumes no cases at the beginning of the time period, a time-dependent method was used to compare results and evaluate if the conclusions were consistent30,31. Briefly, this method has been proposed for estimating the time-dependent reproduction number, Rt, over the course of an epidemic via a likelihood-based simulation method that averages across all transmission networks given a user-specified generation time distribution. The generation time is the lag between infection in a primary case and a secondary case. Because this is often difficult to directly observe, this may be substituted for the serial interval distribution, or the time from symptom onset in a primary case to symptom onset in a secondary case. Ideally, this is estimated from outbreak data using contact-tracing.

The R0 package was used to implement the time-dependent model in R using daily case counts and 10,000 simulations30. Because contact-tracing data were not available, the generation time was parameterized using a Gamma distribution with mean of 7.8 and standard deviation of 4.75, assuming that farms may have approximately a minimum of 3 days between symptom onset (the minimum value of the incubation period, Dlatent) up to approximately 41 days (the maximum value of D estimated from the stochastic doubling-time model plus the maximum latent period of 14 days).

Results

Overview of surveillance

In total, there were 9,422 surveillance reports from January 3, 2023 to March 1, 2024 (Table 1). Of these, 373 (4%) were passive reports and 9,049 (96%) were active reports. When all reports were aggregated by month, the highest number of total reports was in October 2023 (n = 1,028 reports), closely followed by November 2023 (n = 992) and February 2024 (n = 992). Omitting March 2024 (as there are only 3 days of data), the least sampled month was April 2023 (n = 448). There was also a noticeable decline in reports in the weeks of 12/26/2023 and 1/2/2024, coinciding with the Christmas and New Year’s holidays. Overall, because of the much larger number of active than passive reports, these trends closely follow changes in the number of active reports (Fig. 1). The approximate number of unique farms sampled was 6,169. This number is an approximate estimation because there was no consistent farm ID and individual farms could not be easily identified or distinguished within the dataset; thus, there may be fewer unique farms than is estimated here.

Table 1 Count of African swine fever surveillance reports (total, active, and passive) from January 2023 to March 2024, cumulative incidence percentage, number of positive reports and percentage of total reports, and the approximate number of unique farms that had at least one surveillance report (positive or negative), aggregated by province and region.
Full size table
Fig. 1
figure 1

Number of African swine fever reports per week in the Dominican Republic from January 2023 to March 2024, considering all reports (black line), only active reports (red line), and only passive reports (blue line).

Full size image

A map of provinces and administrative regions is available in Supplementary Fig. 1. The Southwest region had the highest number of reports (n = 1,944), whereas the South had the lowest (n = 569). These trends were the same when considering the number of active reports. When considering passive surveillance only, the East region had the most passive reports (n = 89) and Southwest had the least (n = 14).

Not all regions were consistently reported on over time (Fig. 2). The Central, East, North, Northeast, and Northwest regions had an increase in reports that coincided with the onset of the active surveillance program in September 2023. The South, Southwest, and Northcentral regions did not exhibit this same trend or not as strongly.

Fig. 2
figure 2

Number of African swine fever reports in the Dominican Republic each week from January 2023 to March 2024, by region.

Full size image

Geographic distribution of positive results and positive rate

From January 2023 to March 2024, there were 231 positive reports (2.45%) and 9,191 negative reports (97.55%). The majority of positive reports came from passive surveillance (n = 162, 70% of all positive reports; n = 69 positive active surveillance reports). Considering only passive surveillance reports, 43% were positive. Considering only active surveillance reports, 0.76% were positive. While it was not differentiated in the received dataset, surveillance program officials reported that all positive active reports came from backyard farms in the active surveillance program (i.e., not from commercial farms submitting samples every 21 days). There were 221 identifiably unique outbreak farms because there were five producers that appeared twice with positive results in the study period. Four of these farms had two positive reports within an approximately two-month period, and one had two positive reports within a one-week period. The highest number of positive reports came from the East region (n = 53 total) and the lowest from Northcentral (n = 10, Table 1). The provinces of Monseñor Nouel and Peravia each had no positive reports. Peravia had 153 reports from approximately 30 unique farms, and Monseñor Nouel had 320 reports from approximately 112 unique farms.

The positive rate ranged from 0.8% in the Northcentral region to 5.79% in the East region (Table 1). Across the study period, the weekly positive rate ranged from 0% (weeks of August 14–27, September 4–10, and September 25-October 1, 2023) to 8.5% (January 9–15, 2023, Fig. 3). January 9–15 included 8 cases from 3 regions and 4 provinces. The three-week period from June 19-July 9 were the next highest weeks, all above 6.45% positive (n = 28 positive reports).

Fig. 3
figure 3

Weekly African swine fever positive rate of surveillance reports (black solid line) and weekly total reports (dashed red line) in the Dominican Republic from January 2023 to March 2024.

Full size image

The overall cumulative incidence for the DR was 1.49% (Table 1; Fig. 4). The highest cumulative incidence was in the Northwest (6.07%), East (3.55%), and South (2.38%) regions. The lowest was in the North (0.73%), Northeast (0.51%), and Northcentral (0.52%) regions.

Fig. 4
figure 4

Cumulative incidence (%) for provinces in the Dominican Republic considering outbreaks of African swine fever from January 2023 to March 2024. Administrative regional boundaries are designated by black, bold lines. ArcGIS Pro version 3.2.219,20.

Full size image

Reproduction ratio

Using the stochastic doubling time model, the median duration of infectiousness was 17.4 days (95% PI: 9.1–23.9). There were six timepoints at which the number of ASF positive reports doubled (td, Fig. 5). The median R0 estimates and their 95% PI were 4.01 (2.58–5.15), 7.03 (4.16–9.30), 1.55 (1.29–1.75), 1.37 (1.19–1.50), 1.17 (1.09–1.23), and 1.06 (1.03–1.08). From the time-dependent method, Rt fluctuated throughout the time period with no clear trends (Fig. 6). Rt was highest on June 6 (Rt=3.23, 95% PI: 1.0–7.0) and June 9, 2023 (Rt=3.20, 95% PI:1.5–5.0).

Fig. 5
figure 5

Median doubling-time reproduction ratio estimates (black points) and 95% lower and upper limits of the probability interval (red points) for African swine fever outbreaks in the Dominican Republic from November 2022 to March 2024.

Full size image
Fig. 6
figure 6

Median time-dependent reproduction ratio (Rt) estimates (blue line) and 95% lower and upper probability interval (gray shaded region) for African swine fever outbreaks in the Dominican Republic from November 2022 to March 2024. Red, dotted line indicates Rt=1.

Full size image

Discussion

ASF continues to be a significant disease threat globally. This update provides a report on the current status of ASF in the Dominican Republic that can be used by the veterinary authorities, epidemiologists, and private swine stakeholders to develop long-term ASF control programs that are tailored to the country’s current epidemiological situation.

Overall, the vast majority of positive reports came from passive surveillance. This is consistent with passive reports originating from farms with suspicious clinical signs. In 43% of passive surveillance reports, ASF was confirmed. In contrast, active surveillance focused on sampling farms without clinical signs, and only 69 (0.76%) active surveillance reports were positive. These may be farms in the earliest stages of ASF introduction and with too few clinically sick pigs to be noticed by personnel, or with infections presenting with more mild clinical signs, the latter of which has previously been reported in the DR6,32. Additionally, surveillance program staff indicated these positive active reports were all from the active backyard surveillance program and not from commercial farms. It should be noted, however, that samples sent from commercial farms every 21 days are collected and submitted by producers without on-farm oversight by officials16. Some outbreaks may also have been undetected because producers send pigs directly to slaughter without making a passive report. Overall, the low number of positive cases from active reports is expected. Previous studies have indicated that random sampling of apparently healthy animals for ASF surveillance is not effective for early detection compared to passive surveillance or targeted sampling of sick and dead pigs33,34,35. Thus, while the active surveillance conducted here may provide an updated picture of the country as a whole, it is not recommended for long-term monitoring. Regular sampling of sick and dead pigs from farms would likely be more efficient for early ASF detection.

Outbreaks occurred with relative consistency throughout the time period and were geographically widespread across administrative units (region and province). The middle part of the country consisting of the Northcentral, Northeast, and North regions had a notably lower cumulative incidence than surrounding regions (Fig. 4). According to the 2022 census, these regions contain the majority of individual producers (approximately 56% based on the 2022 census) and pigs (52%) in the country10. North and Northcentral also contain over half of the country’s tecnificado or “technified” producers, those commercial farms with the highest level of biosecurity10. With more commercialized producers, these regions may have more farms that implement stricter biosecurity and control measures, ultimately reducing ASF incidence. Notably, the provinces of Peravia (Central region) and Monseñor Nouel (Northcentral region) had no positive reports over the past year. It would be highly beneficial for future work to characterize the situation in these provinces and identify what swine farms may be doing there to remain ASF-free. Throughout the time period, the weekly positive rate remained under 9% without any clear trends (Fig. 3). The increased positive rate in June and July 2023 included outbreaks from all eight regions and, interestingly, coincided with an increase in the total number of passive reports (Fig. 1). This could suggest that there was an underlying behavioral reason or incentive for increased passive reporting of cases, such as an increase in the price of indemnity for depopulated herds compared to market price, or could reflect changes in virus circulation.

Available data from this and a previous analysis13 was also used to estimate country and farm-level R0/Rt from November 2022 to March 2024 (Fig. 5 and 6). The results from both the doubling time and time-dependent methods suggest that ASF’s ecological behavior resembles an endemic state, and in practical terms, that on average an infected farm has already infected another farm by the time it is detected and depopulated. Additionally, the time-dependent method (Fig. 6) displays considerable variability of Rt across time, indicating that the number of outbreaks will likely still be affected by cyclical or seasonal patterns where present. However, these analyses were limited by the assumptions made, including being in an exponential growth period, having no outbreaks at the start of the time period, and assuming a duration of infectiousness for the doubling time method, and assuming a generation time for the time-dependent method. Future data collection from outbreaks should include recording accurate spatiotemporal coordinates, dates of symptom/mortality onsets, and contact tracing. Over time, analyses that are able to incorporate transmission modeling, geographic coordinates, or phylogenetic data may allow for a more precise estimation of R015,36, which could indicate the efficacy of applied control measures.

According to previous official reports, the country’s positive rate of diagnostic testing fell from over 40% in early 2022 to 10.5% in early 202337,38,39. In the current analysis, the positive rate was again decreased, estimated as 2.45% from January 2023-March 2024. However, these positive rates are not easily comparable. From January 2023-March 2024, nearly all reports (96%) came from active surveillance, and when considering passive reports only, the positive rate was 43%. The decrease in the overall positive rate is likely greatly influenced by the large increase in sampling of farms without suspicion or clinical signs through the active surveillance program and the requirement for commercial farms to submit samples every 21 days.

There were important limitations for this work. Overall, spatiotemporal trends were difficult to assess in part because active surveillance was not consistently conducted across the country at the same time (Fig. 2). Because of limited resources and personnel, geographic areas were visited at different times during the approximately 6-month period that the active surveillance program has been in progress. Additionally, as previously stated, both methods of estimating the reproductive ratio included many assumptions due to data limitations.

Nearly three years after the reintroduction of ASF to the DR, these results may suggest the need to change from an approach of emergency response to one of sustained and progressive control. This situation draws many parallels to the ASF situation in Sardinia, where Genotype 1 virus was been present from 1978 to 2024 40,41,42,43. In Sardinia, ASF Genotype 1 was thought to be maintained by complex interactions between illegal free-ranging pigs, backyard farms, and wild boar. In the DR, ASF may be similarly maintained in specific subpopulations of swine farms, such as backyard or –free-ranging farms with poor biosecurity. Additionally, social conditions and limitations in the availability of human and financial resources may have prevented the full enforcement of control activities including controlling movements. This may make swift eradication highly difficult and instead may require a different approach to managing ASF in these populations, such as a combination of official registration, awareness and education, and biosecurity improvements like fencing. In a previous analysis of alternative ASF strategies in the DR, stakeholders highlighted the need to improve ASF control, such as formal registries of pig producers including for small and subsistence farmers, improved infrastructure, increased participation from the private sector, and more public-private partnerships44.

A progressive control plan, such as previously used for reducing foot-and-mouth disease risk in endemic countries45,46,47, may provide a sustainable pathway to build this capacity of the DR public and private sectors and reduce ASF’s impact over time. Registration of both commercial and smallholder swine premises and live pig identification is needed for the public veterinary services to implement effective control strategies, and it would allow for the implementation of more advanced traceability systems. Unique registration identification would also greatly aid epidemiological monitoring and assessment, which was difficult here given the inconsistent farm names in the data. Training programs for swine veterinarians and producers, including specific training tailored to backyard farmers, can help raise awareness and increase the capacity of the private sector to implement biosecurity measures, identify ASF, and potentially increase compliance with reporting procedures. Previous work has been done with smallholders in southeast Asia, and a similar approach could be beneficial here48. Completion of registration and training programs could be paired with specific funding for biosecurity and infrastructure improvements to help incentivize participation in ASF control strategies. Appropriately trained and accredited private veterinarians could support the public veterinary services in specific tasks, such as surveillance and diagnostic testing, particularly because public resources are often limited. Increased funding and training for public veterinary staff would help increase their capacity for ASF response. After initial capacities are built, strategies such as zoning and compartmentalization could be applied to protect disease free geographic areas and commercial systems.

Vaccine use is also an important consideration. As of writing, ASF vaccines are not yet approved for use in the DR, though initial trials have been conducted. If currently available ASF vaccines are eventually approved, they should only be administered by accredited veterinarians following the vaccine manufacturer’s instructions and with appropriate monitoring by the public veterinary services. Monitoring should be supported by premise registration. Further safety, efficacy, modeling, and cost-benefit analysis studies should be performed to understand the optimal use of the vaccine to support ASF control. As a progressive plan is implemented, the public veterinary services may consider mandating vaccine use in certain zones (vaccination zones) as a part of risk reduction efforts and to protect ASF-free zones. Ultimately, building response capacity and reducing ASF prevalence over time may position the DR to eventually eliminate the disease.

In summary, these results provide an evaluation of surveillance activities in the Dominican Republic from 2023 to 2024. ASF incidence appears to have decreased over time, though outbreaks are still being detected by both active and passive surveillance throughout the country. Management and control strategies designed for progressive control may support a long-term goal of ASF eradication in the DR.