Using Retrospective Data to Implement a Mortality Surveillance System for Growing Pigs

Abstract Introduction With survivability being a predominant indicator of success when assessing a production system’s ability to profit, it is no surprise that mortality is at the forefront of industry concern. Mil-Homens et al. (2025) demonstrated that monitoring ongoing weekly mortality data from closeout groups can serve as an early warning system for disease outbreaks in growing pigs. This allows pigs to be treated quicker, enabling more of the herd to survive an outbreak. Using benchmark curves and trained data, we can identify trends and high-risk groups before they break. Objective To implement and evaluate a benchmark-based mortality surveillance system using retrospective nursery data to detect abnormal mortality patterns in lots with high nursery mortality. Materials and Methods This retrospective study utilized weekly mortality data from 6,261 lots, which served as reference lots for the benchmark curve (i.e., the average weekly mortality for these groups was used as the baseline for “normal” mortality). These lots did not experience cumulative nursery mortality above 5%. Standard deviation (S.D.) was used to create threshold zones to trigger warnings set at three levels—1 S.D., 2 S.D., and 3 S.D. Thereafter, 274 lots with known health issues and mortality above 10% were selected as “case” lots and compared to the benchmark curve. The “Break Detection” week was defined for “case” lots when weekly mortality doubled from the previous week. Alerts were then created when a spike in mortality was above the threshold at the three different S.D. levels. Lastly, Early Detection Rates (EDR) were calculated by taking the alert above S.D. levels and “Break Detection” differences within lot by week. Results For the 10% mortality groups (n = 274), at 1 S.D. we saw an average EDR of 2.2 weeks, at 2 S.D. an average EDR of 1.9 weeks, and at 3 S.D. an average EDR of 1.6 weeks. This can also be interpreted as, for level 1 we can detect the break 2.2 weeks prior to on-farm detection. The figure on the right shows the number of lots detected early by week at the respective S.D. levels. Discussion & Conclusion It was determined that the three stipulations to determine the “Break Detection” date provided the conditions for most skilled caretakers to identify an on-farm health event. Using the 1 S.D. warning would give a farmer on average 2.2 weeks’ notice for a health event. Mortality surveillance in growing pigs provides an early warning system for producers that can be used for early intervention or disease monitoring. With added warning, producers can save more pigs and reduce financial loss by implementing treatment plans in a timelier fashion.