Sensitivity, specificity and level of agreement of meat inspector detection of abnormal lymph nodes of farmed deer (Cervus elaphus) in New Zealand

Abstract

AIMS: To assess the sensitivity (Se), specificity (Sp) and level of agreement in the detection of abnormal deer mesenteric lymph nodes (MLN) by official assessors (meat inspectors) for the purpose of slaughter premise surveillance for Johne's disease in New Zealand farmed deer. METHODS: To determine Se and Sp, four meat inspectors in two commercial deer slaughter premises (DSP) each examined between 153 and 925 MLN under normal visual examination, palpation and/or incision inspection procedures. Abnormal MLN were independently defined as those with a circumference measurement of ≥ 55 mm and/or grossly visible pathological changes such as caseation, necrosis and/or mineralisation, while those with a circumference of <55 mm and without gross lesions were defined as normal. The inspectors' diagnosis of normal or abnormal was retrieved from the national DSP-based surveillance database for Johne's disease in deer. To determine between-inspector level of agreement, 54 meat inspectors visually examined two photographic images each of 19 normal and 10 abnormal MLN, and recorded their diagnosis. Between-inspector agreement was calculated based on a modification of Cohen's kappa statistic for multiple raters. The influence of covariates such as inspector age and experience on the between-inspector level of agreement were assessed using the Breslow-Day statistic and test for equal kappa coefficients. RESULTS: The weighted average Se and Sp of inspector detection of abnormal MLN was 13.3 (minimum 4.8, maximum 41.2)% and 99.9 (minimum 99.5, maximum 100.0)%, respectively. The level of between-inspector agreement in the diagnosis of abnormal and normal MLN was fair (κ = 0.32). Employment location, inspector age, experience inspecting deer or other species, and the number of shifts inspecting deer within the previous four weeks had no influence on between-inspector agreement (p > 0.10). CONCLUSIONS: Inspectors diagnosed abnormal deer MLN with a high specificity, but low sensitivity. These data supported that visual assessment of MLN characteristics was suitable for national surveillance for paratuberculosis, while highlighting the need for further emphasis on training of inspectors in abnormal lymph node detection.