Assessment of the use of gross lesions at post-mortem to detect outbreaks of classical swine fever outbreaks

The performance of pathological findings as a diagnostic tool for the detection of classical swine fever (CSF) outbreaks during the 1997/1998 CSF-epidemic in The Netherlands was evaluated by constructing and analysing receiver operating characteristic (ROC) curves. This was done at the individual pig level and at the submission level (a group of pigs from the same herd submitted together for post-mortem investigation). At post-mortem examination, the tonsils, spleen, ileo-caecal valve and renal pelvis were sampled, sent to the reference laboratory, and tested by means of a CSF-specific fluorescent antibody test in combination with a confirmatory test. This resulted in an infection status at the individual pig level. The infection status and pathological findings of 1072 individual pigs from a total of 230 infected herds were included in this analysis. We also included submissions of pigs from herds that were sent to post-mortem examination because of a clinically CSF-suspect situation but afterwards were concluded to be from non-infected herds. Infection status and pathological findings of 1224 individual pigs from a total of 241 non-infected herds were included in the analysis. Pneumonia, pleuritis, chronic bronchitis, pulmonary oedema, chronic gastric ulceration, dry faecal contents in the colon, conjunctivitis, haemorrhages in the renal pelvis, renal haemorrhages, splenic enlargement, haemorrhages in the urinary bladder, haemorrhagic and enlarged lymph nodes were the most frequently recorded pathological findings during a post-mortem examination of pigs submitted in a CSF-suspect clinical situation. However, some of these pathological findings (e.g. pneumonia, pleuritis) were almost evenly distributed in infected and in non-infected pigs, resulting in a high sensitivity combined with a low specificity. The area under the ROC curve of pathological findings at the individual pig level and at the submission level was 0.720 and 0.782, respectively, which was significantly (P<0.0001) larger than the area under the random ROC curve. It was concluded that, although gross pathology is a legitimate test, its quantitative contribution to the detection of CSF is limited.