Part 2: Active sampling
As the focus was on active detection of HPAI virus, sampling was targeted and risk-based, and did not aim to achieve a representative sample. Nomadic ducks were undersampled during the first round, and proportionally oversampled during the second round. No village poultry were actively sampled during the first round; the incentivised passive surveillance was unsuccessful in locating cases or focal outbreaks and was discontinued. Modified snowball sampling was used to perform risk-based sampling in the second round, but the number of samples taken was comparatively low.
Overall, the number of HPAI virus isolates identified was relatively small. H5 virus was detected in all production systems. The differences in detection of AIV and H5 virus between enterprise types and between the two rounds were found to be statistically significant, with significantly more AIV detections during the second round. It could not be determined whether this was a result of differences in the sensitivities of detection between the two rounds, or the consequence of an underlying causal trend. As a consequence of the study design, the scope for model-based analytic approaches was limited and the analysis was mostly descriptive; statistical testing was restricted to hypothesis testing to investigate strengths of association and differences between the enterprise types.
Only the commercial farm sector implemented biosecurity practices and vaccination; however, the extent of implementation was variable between farms and there was evidence that H5 virus was present, albeit at an extremely low level. “Silent spread” and persistence of HPAIV in sub-optimally vaccinated commercial flocks is a concern. In a comparative trial, Tarigan et al. (2018) used seven different commercial vaccines produced in Indonesia mostly based on the H5N1 subtype; vaccination led to highly variable outcomes, including vaccination failures, and was largely ineffective in providing long-lasting protective immunity. Poetri et al. (2014) performed an inoculation trial and showed that a single vaccination applied under field conditions induced clinical protection but was insufficient to induce protection against virus transmission. Given reported movements of birds between Sectors 3 and 4, there is a potential risk of ongoing introduction of H5N1 in commercial flocks. Furthermore, in contrast to the essentially local nature of poultry trade patterns of LBMs and collector yards, the commercial farm network was characterised by longer range movements, which had the potential to spread virus furthest geographically. However, as such movements were mainly to slaughter, this would be unlikely to result in long-distance spatial spread of HPAI.
Snowball sampling was performed to maximise the likelihood of detection of H5 in backyard and nomadic ducks flocks. Such approaches have been successfully used in Cambodia (Poolkhet et al., 2016) and Vietnam (Delabouglise et al., 2016). Despite this, very few H5 samples were detected in the backyard sector and in nomadic ducks. It was evident that underreporting of clinical cases is an issue. Our results indicated that although present in smallholder production systems, the prevalence of AIV appears to be low. Consequently, control measures focusing on this sector may target reporting behaviours, but any direct disease control measures are not indicated.
It was apparent that the largest numbers as well as the highest prevalences of AI, H5 as well as H9 viruses were found in the value and marketing chain – at LBMs and collector yards. This was supported by the pairwise statistical tests for significance between enterprise types. These showed that the odds of detecting H5 were similar in LBMs and collector yards and not statistically different; however, the likelihood of detection at these locations was 3½ to 4 times higher than for backyard birds and nomadic ducks (these differences were marginally to highly significant) and 25 to 30 times higher than for commercial poultry (these differences were highly significant). This finding is consistent with results elsewhere in South and South East Asia. It could not be determined from our results whether the value chain concentrates or even amplifies virus along its length, or whether the LBMs and collector yards are locations in which AIV persists and actively circulates (or, indeed, a combination of both). Our SNA suggested that LBMs represented high-risk enterprises due to their small world network properties which would enable disease spread more rapidly. Collector yards played a role as a bridge between enterprises spreading the virus. Commercial farms had a possible role in the long-range spread of HPAI virus.
The pattern for H9 was somewhat different. More samples from commercial farms tested positive than would be expected; approximately half of all the AIV detected were H9. Equally surprisingly, relatively few H9 were identified from nomadic ducks. Overall, H9 were identified in 10/41 collector yards (25%), 10/18 LBMs (55%) and 5/31 commercial farms (16%) over the duration of the study. A number of collector yards as well as LBMs repeatedly tested positive. One LBM in particular (M-12) was positive at every sampling. No commercial farm tested positive more than once; however, these farms were only sampled half as frequently. As the sampling no or limited repeat sampling was carried out in backyard and nomadic duck flocks, we were unable to assess this in these sectors. Therefore, although there was some variation, the H5 and H9 viruses were present year-round and were co-circulating, although no inference can be drawn regarding any interactions between these HPAI and LPAI viruses.
It is likely that this higher rate of detections represents a potential public health risk. For example, in one study (Shimizu et al., 2016) 85 of 101 sera that were collected from workers at a LBM in East Java tested positive for AIV antibody; over a two-year period, the average HI titre nearly trebled, and seroconversion occurred in 11 of 25 sera (44%). However, a questionnaire indicated that no workers had experienced an episode of severe acute respiratory illness.