Highly pathogenic avian influenza subtype H5Nx clade 2.3.4.4 outbreaks in Dutch poultry farms, 2014–2018: Clinical signs and mortality

Abstract In recent years, different subtypes of highly pathogenic avian influenza (HPAI) viruses caused outbreaks in several poultry types worldwide. Early detection of HPAI virus infection is crucial to reduce virus spread. Previously, the use of a mortality ratio threshold to expedite notification of suspicion in layer farms was proposed. The purpose of this study was to describe the clinical signs reported in the early stages of HPAI H5N8 and H5N6 outbreaks on chicken and Pekin duck farms between 2014 and 2018 in the Netherlands and compare them with the onset of an increased mortality ratio (MR). Data on daily mortality and clinical signs from nine egg‐producing chicken farms and seven Pekin duck farms infected with HPAI H5N8 (2014 and 2016) and H5N6 (2017–2018) in the Netherlands were analysed. In 12 out of 15 outbreaks for which a MR was available, MR increase preceded or coincided with the first observation of clinical signs by the farmer. In one chicken and two Pekin duck outbreaks, clinical signs were observed prior to MR increase. On all farms, veterinarians observed clinical signs of general disease. Nervous or locomotor signs were reported in all Pekin duck outbreaks, but only in two chicken outbreaks. Other clinical signs were observed less frequently in both chickens and Pekin ducks. Compared to veterinarians, farmers observed and reported clinical signs, especially respiratory and gastrointestinal signs, less frequently. This case series suggests that a MR with a set threshold could be an objective parameter to detect HPAI infection on chicken and Pekin duck farms at an early stage. Observation of clinical signs may provide additional indication for farmers and veterinarians for notifying a clinical suspicion of HPAI infection. Further assessment and validation of a MR threshold in Pekin ducks are important as it could serve as an important tool in HPAI surveillance programs.


Clearly, early detection of HPAI virus infection on poultry farms
is essential to reduce risks for virus spread and minimize the socio-economic impact of the disease (Backer, van Roermund, Fischer, van Asseldonk, & Bergevoet, 2015;Elbers, Fabri, et al., 2004), which is also increasingly reflected in legislation and contingency plans worldwide. European Union legislation on the control of HPAI (EU, 2005a(EU, , 2005b stipulates that early detection systems, aimed at a rapid reporting of any sign of avian influenza in poultry and other captive birds by owners or keepers to the competent veterinary authority, need to be in place. For both LPAI and HPAI outbreaks, sudden changes in mortality have shown to be an indicator of infection (Elbers, Holtslag, Bouma, & Koch, 2007;Gonzales & Elbers, 2018;Malladi, Weaver, Clouse, Bjork, & Trampel, 2011), as well as clinical signs (Elbers, Kamps, & Koch, 2004;Elbers, Koch, & Bouma, 2005;Velkers et al., 2006).

These indicators have been used to formulate criteria in European
Union legislation for reporting suspicion of a notifiable disease such as avian influenza in poultry, with even more detailed criteria implemented in national regulations in the Netherlands (Box 1). However, the current reporting thresholds may not be sensitive enough for timely detection of HPAI virus infections (Gonzales & Elbers, 2018).
Published reports on analyses of mortality data from previous outbreaks, that is HPAI H7N7 in 2003 (Bos et al., 2007;Stegeman et al., 2004) and HPAI H5N8 in 2014 and 2016 (Velkers, Elbers, Bouwstra, & Stegeman, 2015) have shown that (a) it takes several days after the start of increased mortality due to HPAI virus infections to reach the official reporting threshold of 0.5% mortality for two consecutive days; and (b) many flocks have already been depopulated well before reaching these thresholds. To improve sensitivity of detection of LPAI and HPAI virus infections and at the same time maintain a high level of specificity, Gonzales and Elbers (2018) developed new reporting thresholds based on increased mortality and drops in egg production for layer farms, and evaluated the performance of those indicators with HPAI H7N7 outbreak data from 110 infected layer flocks in the Netherlands in 2003. The mortality ratio (MR), with a reporting threshold of 2.9 times higher mortality than the average weekly mortality of the previous week for that particular flock, had a 95.3% sensitivity to signal HPAI virus infection in laying hens and would have resulted in 2 days earlier detection compared with the current Dutch national thresholds for HPAI and in 7 days earlier detection for LPAI virus infection (Gonzales & Elbers, 2018).
For early detection of HPAI virus infections, the suggested MR ratio threshold of 2.9 may also be applicable to other poultry types. Ssematimba et al. (2019)  Annex II: criteria to be considered when applying the measure set out in Article 2: drop in feed and water intake higher than 20%; drop in egg production higher than 5% for more than two days; mortality rate higher than 3% in a week; and any clinical sign or post-mortem lesion suggesting avian influenza. Poultry keepers have to report increased mortality in layers, reproduction birds or broilers (older than 10 days) to the authorities in case of 0.5% mortality or more per flock per day for two consecutive days; in turkeys in case of 1% mortality or more per day for two consecutive days; and in AI susceptible birds in case of 3% or more mortality per

Dutch Ministerial Regulation
week.
Poultry keepers of AI susceptible birds need to consult their veterinarian in case of a clinical problem; reduction in feed intake or water intake of 5% or more per day for two consecutive days; in layers and breeders a reduction in egg production of 5% or more per day for two consecutive days.  in the Netherlands. We calculated the MR and daily mortality for each outbreak and provide an extensive inventory of the species-specific clinical signs and how these developed over time in the days before official notification, as observed by poultry farmers and veterinarians.  b Diagnosis of HPAI, tested positive on real-time PCR on the matrix gene, H5-PCR and sequencing of the haemagglutinin and neuraminidase (Beerens et al., 2018). c Samples were submitted to the national reference laboratory by the veterinary practitioner in the framework of the Dutch early-warning system, we considered the day of the positive result of these samples as day of notification. d Two flocks infected with HPAI virus present on the farm, one flock age 15 days the other age 43 days. least daily records of mortality and production data, for example feed and water intake, and egg production. All birds on the HPAI virus-positive farms were culled within 1-2 days after the day of notification (Table 1).

| Data collection
Additionally, an in-depth epidemiological investigation was performed by specialized poultry veterinarians of the Faculty of Veterinary Medicine of Utrecht University. This investigation was performed for all farms between 9 days to 3 months after culling and was aimed to facilitate retrospective identification of the most likely moment and route of HPAI virus introduction and/or spread (referred to as Detailed Epidemiological Investigation [DEI]). For all farms, all available data collected by NVWA and laboratory results

F I G U R E 1
Clinical signs observed by the farmers, categorized by organ system, for the highly pathogenic avian influenza virus-infected chicken (left) and duck farms (right) and exceedance of daily mortality (>0.5%) and mortality ratio (MR) thresholds in the 5 day period prior to notification. † Day of notification for D-1 was the day a positive result was found in the early warning swabs sent in by the veterinary practitioner. ‡ Not enough mortality data were available to calculate the mortality ratio notification up to and including the day of the VIV. These data were used for further data analyses as described below.

| Mortality and production parameters
Mortality ratio (MR) and egg production ratio (EPR) were calculated as described by Gonzales and Elbers (2018) for each of the flocks, using available flock records of at least 5 days to approximately 1 month before notification. The threshold of 2.9 for MR, as applied for laying hens by Gonzales and Elbers (2018), was used and the first day the MR exceeded the threshold was considered as an increase in MR and used for further analyses. We were not able to calculate the MR for one Pekin duck farm (D-2) due to incomplete mortality data in the weeks prior to the outbreak. The current applied daily mortality (DM) threshold of 0.5% per flock (see Box 1) was also used for comparisons.
In layer farms, an EPR of below 0.94 was considered as presence of reproduction tract signs. The use of this threshold alone, and in combination with the MR, was validated as a way to detect LPAI and HPAI outbreaks at an early stage by Gonzales and Elbers (2018). Data on daily growth were not recorded in any of the affected farms. In farms where records of water and feed intake were available, a decrease in feed or water intake of 5% compared with the previous day was classified under general clinical signs as described below.

| Clinical signs
The standardized form used to record clinical signs observed during the VIV included a yes or no checklist with questions on feed and water intake, sudden death, ruffled feathers, diarrhoea, egg quality, oedema and cyanosis, nervous signs, abnormal conjunctivae, lacrimation, respiratory distress and decreased activity. Furthermore, the veterinarians recorded findings on mortality, production and feed and water Also, the flock records were checked for notes on clinical signs.
A list of clinical signs, categorized in different categories, was used to compile all the data from the veterinarians from VIV and DEI, and only from de DEI for the farmers separately. The observed clinical signs were categorized in six categories, that is as clinical signs attributed to nervous and locomotor system; mucosal membranes and skin; respiratory tract; gastrointestinal tract; and reproduction tract (Tables S1-S4) or as general clinical signs. The latter category included signs of general illness, which could not be related to a specific organ system or were associated with signs of systemic disease, for example depression, reduced feed or water intake, ruffled feathers or hunched posture, cold or warm extremities and sudden death (Tables S1-S4). Signs of the nervous and locomotor system were categorized together as these were difficult to distinguish based on the information from the farmers.

| Outbreaks
Five, eight and three farms were infected in the autumn-winter pe-

| Clinical signs
A detailed list of the observed clinical signs by farmers and veterinarians in the chicken and duck flocks based on VIV and DEI is provided in Tables S1-S4. Figure 1 summarizes the clinical signs that were observed by farmers in their flocks in the 5 day period prior to notification to the authorities, and occurrences where the current official DM threshold for reporting (>0.5%) or the MR threshold (>2.9) were exceeded.

| Chicken farms
In both parameters, the first day the parameter exceeded its threshold was used in the further analyses.

| Pekin duck farms
In the Pekin duck farms (n = 7 outbreaks), the first clinical signs were observed 2 days prior to the notification in two outbreaks (D-1 and D-5), which included general clinical signs and signs of the nervous or locomotor system and respiratory tract (Figure 1).
For outbreak D-5, temporary sneezing was only observed at day two before notification. A day prior to notification farmers ob-

| Pekin duck farms
The single day and was <2.9 the following day. On one occasion, the MR exceeded the threshold on two consecutive days in one house of a farm (D-4.1). This house also had the most occasions (five out of seven) in which the MR temporarily exceeded the threshold.
The MR exceeded the threshold in three out of six outbreaks prior to observation of clinical signs. In one outbreak, the increase of MR coincided with the first observation of clinical signs, and in two out of six outbreaks the clinical signs were observed prior to an increase in MR (Figure 1).

| D ISCUSS I ON
The purpose of this case series was to describe the observed clini- Moreover, in eight out of nine outbreaks in chicken and Pekin duck farms where the DM did exceed 0.5%, the MR had already exceeded its set thresholds 1-4 days prior. In pheasants, however, it was found that exceeding a set absolute threshold on two consecutive days resulted in the best trade-off between false-alarm rate and early detection compared with a 7 day moving average or exceeding a set absolute threshold for 1 day (Ssematimba et al., 2019). Due to the limited data set, we were not able to evaluate these trade-offs appropriately, but the results obtained from these H5Nx outbreaks in the Netherlands suggest that the MR could be a more sensitive  (Table S1). Early in the flock infection process, however, the farmers in our study mainly observed clinical signs that could only be considered as general clinical signs, which are not specific for HPAI virus infection (Elbers et al., 2007;Swayne, Suarez, & Sims, 2013)  Compared to the veterinarians, farmers observed and reported less specific clinical signs, especially regarding respiratory and gastrointestinal signs in both chicken and ducks. This difference may be due to the specialized training and experience of the veterinarians in poultry veterinary medicine to observe signs of disease, and veterinarians may be better equipped with a repertoire of specific words to indicate their observations and relate that to a specific organ system.
The discrepancy in observation of clinical signs between farmers and veterinarians is, however, smaller than we anticipated, suggesting that the farmers were aware of signs to look for. This shows that training and awareness of the farmer in detecting clinical signs is an important tool in detecting HPAI virus infection at an early stage.
The willingness of the farmer and practitioners to report a suspicion of a notifiable disease to the authorities may be different for the very first suspicion compared with suspicions after the first confirmed HPAI outbreak (Elbers et al., 2010). To prevent the spread of HPAI viruses to other farms, it is crucial to notify a suspicion as early as possible to be able to adequately diagnose and quickly depopulate the farms. The first outbreak of a HPAI (H5Nx; outbreak no. L-1) in 2014 had increased mortality (>2.9) for 5 days prior to notification. In the outbreaks after 2014, the mortality ratios exceeded the threshold 0-3 days prior to notification, which suggest that farmers were more alert and reported a suspicion of notifiable disease more rapidly. Additionally, two Pekin duck farms and one chicken farm had multiple outbreaks of HPAI in 2014, 2016-2017 on their farms, making the farmers even more aware of the risk of a new outbreak. Due to the fast reporting of HPAI suspicion of farmers and veterinarians to the authorities, the spread of HPAI viruses to other poultry facilities was minimized.
To conclude, the current study gives an indication that the use of an objective MR with a set threshold could be a reliable parameter to detect HPAI virus infection on chicken and Pekin duck farms at an early stage and may perform even better when complemented with detection of clinical signs in poultry farms, provided farmers are well trained to notice them. These results underline the need to validate the MR in Pekin ducks and other poultry species, and it should encourage farmers, veterinarians and veterinary institutes in other countries to monitor and register mortality on farms more rigorously, because a poultry-specific MR could serve as an important indicator in HPAI poultry surveillance programs.