Validation of producer-recorded causes of preweaning mortality in swine

Abstract

This study examined the accuracy of producer-recorded causes of preweaning mortality on 13 commercial swine farms using a computerized recording system. All stillborn piglets and those born alive but dying between birth and weaning were collected over 6–8-week period and stored frozen until necropsy. Reason for death, piglet age, date of death and antemortem history, when known, were recorded by producers for 1063 dead piglets. There were 916 piglets born alive but dying before weaning, 146 stillborn piglets and one mummified piglet. Gross observations at postmordem examination were used to classify piglets into the following major categories of mortality: low viability, trauma, scours, deformed, infection, stillborn, and other. The relationship between producer-recorded causes of death and causes established at necropsy was expressed in terms of sensitivity and specificity. Sensitivity was the proportion of piglets correctly classified within a category, using necropsy results as the gold standard. Specificity was the proportion of piglets correctly excluded from a category. The sensitivity and specificity of producer assessment of the death causes varied with the mortality category. Within each category, variations in sensitivity and specificity also were noted among age groups. The results demonstrated that without validation of data-recording accuracy, retrospective and prospective studies using producer diagnosis of causes of death in suckling piglets are likely to be invalid.