Bayesian Framework of the sensitivity and specificity of five pneumonia detection methods in pre weaning calves

• Ano de defesa: 2021
• Autor(a) principal: Decaris, Nathália
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
• Palavras-chave em Português: Accuracy; Acurácia; BRD; DRB; Escores; Haptoglobin; Haptoglobina; Scores; Ultrasonography; Ultrassonografia
• Link de acesso: https://www.teses.usp.br/teses/disponiveis/10/10136/tde-16032022-092500/

Calves ′ bronchopneumonia is a multifactorial disease, which causes short- and long-term economic losses, being early detection a crucial factor in relation to therapeutical response, disease severity and mortality. Due these factors clinical scores have been developed in North America for screening sick animals at farm level, as well as the use of ultrasonographic scanning to detect consolidation lesions in lung parenchyma. Over the environmental conditions and raising system, the diagnostic methods developed at North America has been applied in subtropical and tropical countries. Due theses facts, this research evaluated the accuracy of five methods to diagnose bronchopneumonia in Holstein dairy calves at different raising system in subtropical conditions. The tests used to diagnose bronchopneumonia were thoracic ultrasound (TUS, positive if consolidation depth ≥1cm), auscultation (Ausc, positive in presence of crackles, wheezes and pulmonary silence areas), Wisconsin score (Wisc, positive if ≥4), Californian score (Calif, positive if ≥5) and serum haptoglobin concentration (Hap, positive if ≥15 mg/dL) for BRD diagnosis. Heifers with 30 days to weaning age (n= 482) of 17 commercial farms in São Paulo state were enrolled at winter, spring and summer seasons. Environmental data (Vmax, Vmed, Temperature, Chill, Dew Point, Humidity, Heat Index) was assessed in each farm in sampling and was compared between winter and spring/summer. Bayesian latent class models were used with informative priors for TUS, AUSC and Hap accuracy (sensitivity (Se) and specificity (Sp)) and non-informative priors for Wisc and Calif accuracies. Environmental data showed that São Paulo climate was predominantly hot and humid in the two seasons, and differs only at humidity levels presented difference in the middle day (p=0.03) and afternoon (p <0.01) between spring/summer and winter season, and dew point was higher in middle day (p=0.01) and afternoon (p <0.01) at summer/spring, no differences was observed at temperature levels between seasons. Percentual of positives calves to bronchopneumonia for Wisc score vary from 40 to 80%, Calif escore was 8-51%, TUS 0-72%, Ausc 0-32% and Hap 0-100%. Maximum consolidation depth obtained was 4,99cm. The Se (95% credible interval (CI)) and Sp Wisc were 97.3% (89.499.9) and 52.5% (46.658.8), For Calif Se was 60.0% (47.672.4) and 80.7% (75.8-85.5), TUS presented Se 67.7% (57.277.8) and 87.8% (83.392.1), AUSC Se was 50.7% (37.5-67.2) and Sp was 99.2% (96.9-1), Hap presented Se and Sp of 68.0% (58.7-76.7) and 51.0% (46.1-56.1). Evaluated scores presented inverse accuracy levels, Wisc with high Se and low Sp, and Calif with high Sp and low Se, needing to be adapted by their utilization. No specif test achieved Se and Sp levels to diagnose alone the bronchopneumonia in calves, therefore the use of scores for screening BRD associated to TUS can reduce the inadequate diagnosis.