Ability of veterinary pathologists to diagnose classical swine fever from clinical signs and gross pathological findings.

Clinical signs recorded in a classical swine fever (CSF)-suspect situation and the results of the subsequent post-mortem examination (PME) from swine submitted to post-mortem during the 1997-1998 CSF epidemic in the Netherlands, were presented in an experiment as anonymous cases (without knowledge of the actual infection status of the submission) to five veterinary pathologists for their judgment: CSF-suspect or non-suspect. It was presented to them in two hypothetical situations: country was free of CSF for 5 years and CSF was detected in the country 2 weeks ago. Subsequently, their judgment was compared to the gold standard (infection status of the submission on the basis of an immunofluoresence assay on tissue samples) and the sensitivity (Se) and specificity (Sp) of clinical diagnosis was estimated. Furthermore, intra- and inter-pathologist agreement on pathological diagnosis was measured. Adding information on clinical signs to information on PME resulted in a significant increase in median Se and a significant decrease in median Sp for a clinical judgment by pathologists. Also, median Se was significantly higher-and Sp significantly lower-for a clinical judgment by pathologists in a situation in which CSF was detected 2 weeks ago in the country compared to the situation in which the country had been free of CSF for 5 years. Apparently, the internal threshold of pathologists is severely adjusted depending on the actual disease situation in a country. Intra-rater agreement ranged from fair to almost perfect for a clinical diagnosis on the basis of PME. There was considerable variation between pathologists, especially if clinical diagnosis was based on the combined information on PME and clinical signs observed in the herd. Inter-rater agreement was substantial for the diagnosis based on information on PME. Inter-rater agreement was considerably lower for the diagnosis based on the combined information on PME and clinical signs observed in the herd.