Porcine reproductive and respiratory syndrome virus continues to be the most impactful disease to U.S. swine production. Improving the disease surveillance system is of utmost importance to early detection of a PRRSV outbreak.

Conventional PRRSV surveillance includes routine diagnostic testing for PRRSV nucleic acid (by PCR-based assays) and/or for anti-PRRSV-antibodies (by serology) using biological samples including processing fluids (Lopez et al., 2019; Trevisan et al., 2019), oral fluids, family oral fluids (Prickett et al., 2008; Almeida et al., 2020) or serum samples (Ramirez and Karriker, 2019).

Even though ongoing surveillance based on diagnostics (e.g. weekly processing fluids) is reliable and affordable, in the field once a sow farm reaches stability, field veterinarians change the frequency of the diagnostics and start relying more on key production indicators to identify deviations from productivity. Syndromic surveillance can provide valuable information to swine producers and allows them to do early investigation and system-wide intervention.

PRRSV infection is characterized by substantial changes in the performance, and in the behavior of the animals. The most common changes at the sow farm are increase in the number of abortions, increase in preweaning mortality, reduced feed intake and off-feed events. Thus, the systematic surveillance of weekly production indicators or other data sources allows the detection of deviations in productivity or clinical signs that are closely associated with disease outbreaks and can serve as triggers to further investigation.

The objective of this study was to assess the time to detect PRRSV outbreaks using operational data consolidating multiple data streams. For that, we consolidate weekly production data from recording keeping systems, off-feed events from sow electronic feeding systems and diagnostics from six breeding herds from 2020 to 2022. The exponentially weighted moving average model (figure 1) was used to determine early detection, time to detect disease (time between first alarm raised by the model and the week the farm diagnosed the outbreak) and time to baseline productivity (time between first alarm and last alarm until the parameter returned to the baseline).
 

Figure 1. Exponentially weighted moving average for percentage off-feed events with a baseline of 20 weeks.

Table 1 summarized the preliminary results where the early detection rate was higher for abortions (100%), followed by off-feed events (60%), pre-weaning mortality (50%) and finally dead sows and litters weaned (33.3%). Regarding the time to detect disease, the model was able to detect signs of outbreak up to 5 weeks before the farm diagnosed the PRRSV outbreak for abortions, off-feed events, dead sows and litters weaned, and up to 4 weeks before for pre-weaning mortality.

Abortions and off-feed events returned to baseline after 10 weeks. For dead sows, litters weaned and pre-weaning mortality the parameters did not return to baseline in the observed period for this analysis. Abortions had the best performance with 0% of false negative rate and 0.98% of false positive rate.
 

Table 1. Early detection, time to detect disease and average time to baseline productivity.

This study allows us to think further about the use of different data sources and key production indicators that are collected every day on syndromic surveillance models to monitor disease and to early identify signals associated with disease introduction. If this tool were used weekly or daily, action could be taken weeks earlier compared to when the farm diagnoses the outbreak.

Key takeaways:

Source: Mafalda Pedro Mil-Homens, Daniel Linhares and Gustavo Silva, who are solely responsible for the information provided, and wholly own the information. Informa Business Media and all its subsidiaries are not responsible for any of the content contained in this information asset.