Risk-Based Avian Influenza Surveillance System for Poultry in The Republic of Korea: Response to H5/H7 Virus Detection in Wild Birds

Avian Influenza Virus (AIV) belongs to the alphainfluenzavirus genus of Orthomyxoviridae and is infected with various types of vertebrates such as mammals as well as birds. AIVs are classified into various subtypes based on the combination of the different surface proteins, Hemagglutinin (HA) and Neuraminidase (NA). So far there are 16 HA and 9 NA subtypes, therefore the combination of HA and NA results in 144 subtypes, theoretically [1]. Considering that, in H5 and H7 virus types, cases of highly pathogenic have been confirmed, when either of these two types of AIV is detected, a response must be promptly implemented in regarding as highly pathogenic. In the Republic of Korea, the winter management of Highly Pathogenic Avian Influenza (HPAI) is strengthened by setting special control period from late autumn of October to early spring of February or March. During this period, specimens of wild birds (e.g. feces, dead bodies, captures) are collected and actively surveilled by testing on AIVs [2]. In this study, a comparison will be done, between winters of 2019/2020 and 2020/2021, on the livestock vehicles and farms, which were epidemiologically linked to the detection of H5 AIV in wild birds. In 2020, starting with the first detection on October 10, 23 cases of H5 viruses were detected in wild birds of 17 regions (cities and counties), but HPAI was not confirmed. Meanwhile, in the winter of 2020/2021, a total of 297 cases of H5 or H7 AIV were detected. Of these, 234 cases detected in 65 regions were confirmed to be H5N8 HPAI virus. The first detection was confirmed on October 21, 2020. In addition, HPAI was confirmed in 109 poultry farms in 48 regions. And in 25 regions, HPAI was confirmed in both wild birds and poultry.
The number of H5 AIV detection in wild birds were 8 (34.8%) in October and November, respectively, 3 (13.0%) in December, and 2 (8.7%) in January and March, respectively in 2019/2020. Considering that the largest number of H5 AIV detection was recorded in October and November, when most movement of winter birds into Korea are done, detection of H5 AIV is anticipated to be decreased from December, if its transmission is minimized in the groups of wild birds. However, a different pattern of detection was observed for H5N8 HPAI virus. The number of detections in wild birds were 2 (0.9%) in October, 13 (5.6%) in November, 46 (19.7%) in December, 121 (51.7%) in January, 42 (17.9%) in February, and 10 (4.3%) in March. The complex of favorable environmental condition of the virus survival (i.e. the lowest temperature of the year in January) and the spread within the wild bird population seems result in the highest number of HPAI virus detection in January. The number of HPAI outbreaks in poultry was also the highest in December and January with 41 (37.6%) each in December and January. One (0.9%) each in November and April, 20 (18.3%) in February (18.3%) and 5 (4.6%) in March were confirmed. A risk assessment system was developed to give information on the risk of HPAI on poultry farms in relation to a detection of H5/H7 AIV in wild birds. From the day before the collection of specimens detecting H5 AIV to the day of confirming the detection through laboratory test, livestock vehicles that passed a radius of 3km from the sampling point were selected and their visits to poultry farms were tracked. In this system, data on the trajectory of movement emitting from GPS devices attached to livestock vehicles were used [3]. As soon as H5 AIV detection is confirmed in wild bird specimens, the risk of HPAI was assessed for each region where poultry farms visited by these livestock vehicles are located. The regions were classified into four risk groups. In this risk assessment, deep learning techniques for farm level estimation, Gaussian Mixture model for region level estimation, and K-means method for risk group classification were sequentially applied [4]. Risk was calculated for 151 regions in relation to 23 detections of H5 AIV in 2019/2020, and for 163 regions in relation to 234 detections of H5N8 HPAI in 2020/2021. The regions in 2020/2021 includes all 48 regions where HPAI outbreaks have been confirmed on poultry farms. Number of regions which were linked to the AIV detection in wild birds through livestock vehicles were from 3 to115 (Minimum, Maximum, Median 53) in 2019/2020, and 1-145 (Median 46) in 2020/2021. Most regions linked with the AIV detection points were located in the central part of the country, including southern part of Gyeonggi-do, northern Chungcheongnam-do, and northern Chungcheongbuk-do, having lots of traffic. Large numbers of regions were linked when AIV was detected in Anseong (Gyeonggido), Asan and Cheonan (Chungcheongnam-do), and Cheongju (Chungcheongbuk-do) in 2019/2020. Similarly, in 2020/2021, they were Cheonan, Icheon (Gyeonggi-do), and Cheongju. Since the risk at regional level was independently assessed for every detection of AIV, a specific region might be the subject of estimation for many times. The regions for which the risk of HPAI were estimated most frequently were Cheongyang (Chungcheongnam-do, 17 times), Gongju (Chungcheongnam-do) and Sangju (Gyeongsangbuk-do, 16 times each) in 2019/2020. In 2020/2021, they were Pocheon (Gyeonggi-do,141 times), Cheonan (133 times), and Anseong (132 times). This means that vehicles that have passed through the points of detecting wild bird AIV visited farms in those regions the most. When the risk was weighted by these frequencies the overall risk was the highest for Sangju (46 scores) followed by Cheonan and Asan (44 scores each) in 2019/2020. In the case of 2020/2021, it was the order of Pocheon (430 scores), Icheon (404), Hwaseong (Gyeonggi-do) and Dangjin (Chungcheongnam-do) (391 each), and Cheonan (388). In the case of poultry, the largest number of HPAI outbreak was confirmed in Icheon (7 farms). And lots of outbreaks were also confirmed at poultry farms in areas where high risks have been predicted in association with HPAI virus detection in wild birds. HPAI was confirmed in 6 farms in Cheonan, 4 each in Pocheon and Hwaseong.
In 2019/2020, after passing 23 points of detecting H5 AIV, 4,144 livestock vehicles visited poultry farms, and they visited 7,099 farms for a total of 19,900 times. According to arithmetical calculation, one vehicle visited farms for 4.8 times. In 2020/2021, 16,580 livestock vehicles visited 26,635 poultry farms after passing 234 points of detecting H5N8 HPAI virus. The total frequency of farm visits was 288,333, with 17.4 visits per vehicle. As for the number of vehicles by type, animal transport was the largest in both 2019/2020 and 2020/2021, followed by feed lorry, consultant, and egg transport. However, the frequency of farm visits per livestock vehicle varied depending on the year. In 2019/2020, the number of movements per vehicle was the largest for egg tray transport (9.7 times). The second and the third were animal manure treatment (5.8) and veterinary pharmaceutic transport (5.6). In 2020/2021, feed lorry (25.8 times), egg tray transport (24.1), and animal transport (17.7) moved frequently. The reason why the frequency of farm visits per vehicle increased in 2020/2021 compared to 2019/2020 is interpreted as an association with the increase in the number of detections of AIV. One vehicle might be related with several detections. It showed the necessity of intensive management on the vehicles which are numerous or frequently moving, including animal transport, feed lorry, consultant, animal manure treatment, and egg tray transport. The results of this study showed that poultry farms with HPAI outbreak were observed in the regions with prediction for high risk. For example, both in 2019/2020 and 2020/2021, Cheonan was predicted as high risk and the third largest number of farms with HPAI outbreak was confirmed in this region. In this study, risk assessment at region level was based on deep learning technique, applying multi-layer perceptron and backpropagation, which has the characteristics of a black box. Therefore, it is a restriction that the estimated risk could not be clearly explained [5].
The risk of HPAI for poultry farms was estimated for each detection of AIV in wild birds. And the assessed risk was notified to animal health agencies with lists of vehicles, farms, and their links by movement of vehicles. In the near future, indicators that can objectively understand risk factors (i.e. animal husbandry and environment of poultry, wild bird habitats, AIV detection status, and timing requirements, etc.) should be added.

This work was conducted with support from the Animal and Plant Quarantine Agency [Research project numbers B-1543068- 2018-19 & Z-1543068-2018-19].

1. (2020) International Committee on Taxonomy of Viruses.
2. (2021) Ministry of Agriculture, Food and Rural Affairs. Instructions on Control of Avian Influenza.
3. (2021) Animal and Plant Quarantine Agency. Korea Animal Health Integrated System.
4. Yoon H, Lee I, Cho G, Kim H, Lee E (2021) Surveillance of highly pathogenic avian influenza on poultry farms in tracking livestock vehicles in the Republic of Korea. Appro Poult Dairy & Vet Sci 8(4).
5. Yoon H, Jang A, Jung C, Ko H, Lee KN, et al. (2020) Risk assessment program of highly pathogenic avian influenza with deep learning algorithm. Osong Public Health Res Perspect 11(4): 239-244.