B.FLT.3010 - Determination of bovine respiratory disease diagnostic accuracy for multiple modalities

Summary

Accurate and early disease diagnosis at feedlots is important to improve feedlot performance, health, welfare and carcase characteristics. Bovine Respiratory Disease (BRD) results from the complex interaction of pathogens with animal, environmental and management risk factors. Human observation combined with current diagnostic technologies have only a moderate sensitivity and specificity for diagnosis.

This project determined the accuracy (sensitivity, Se; specificity, Sp) for a range of currently available and developing BRD diagnostic tests relative to the gold standard of pulmonary lesions at the time of diagnosis. Diagnostic modalities included ultrasound, haptoglobin, computer aided auscultation, nasopharyngeal microbiota, metabolomics) and continuous monitoring (behaviour collar tags for the Remote Early Disease Identification (REDI) system, reticulo-rumen thermobolus temperature monitoring).

Overall thoracic ultrasound, metabolomics, haptoglobin, reticulo-rumen boluses and REDI all showed promise at identifying pulmonary lesions in animals with 3 consecutive clinical illness scores over a 24-hour period and rectal temperature over 40 C. Further research is required to determine the practical utility and accuracy of these screening and confirmatory diagnostics in field settings.

Objectives

(1) Determine diagnostic accuracy and ability to delineate disease severity with BRD screening tests (e.g. thoracic ultrasonography, thoracic auscultation, acute phase proteins, respiratory microbiota, metabolomics).
(2) Determine diagnostic accuracy and ability to delineate disease severity using confirmatory tests (e.g. thoracic ultrasonography, acute phase proteins, respiratory microbiota, metabolomics).
(3) Estimated ability of continuous monitoring systems for early identification and delineating BRD severity (e.g. rumen temperature bolus and Remote Early Disease Identification, REDI, system).
(4) Comparison and contrasting of potential combinations of diagnostic modalities into a scheme to optimise detection and confirmation of BRD cases.

Key findings

• The project was conducted using two replicates (March 2020, March 2021, n = 200, 190 respectively) of commercially sourced feeder cattle monitored at a collaborating feedlot. Based on visual clinical illness symptoms, rectal temperature and necropsy confirmation, 10 x CASE and 10 x CONTROL (CONT) animals were identified during the feeding phase.
• For CASE and CONT, the best models for each modality examined resulted in ultrasound reporting as the most accurate (Se: 100%, Sp: 90%), followed by plasma haptoglobin (Se: 100%, Sp: 78%) and metabolomics (model random forest for visual/clinical signs from B.FLT.0164; Se: 100%, Sp: 78%), reticulo-rumen bolus (model classification tree T-180; Se: 100%, Sp: 70%), the REDI system (model REDI_ds3; Se: 90%, Sp: 80%), and computer aided auscultation (Se: 60%, Sp: 60%).
• A combination of visual observation (3 consecutive clinical illness scores => 2 over a 24 period) and a rectal temperature of greater than 40 C had a Se 91% and Sp 77%.
• There were multiple study feeder animals scored with CIS => 2 at least once, but not 3 consecutive times per the study protocol; therefore, these cattle did not qualify as a CASE. Of the 29 called abnormal (respiratory, CIS => 2) at least once by visual observation, not enrolled as CASE animals, not treated for BRD per the feedlot treatment protocol, and ultimately managed to harvest slaughter, only 3 presented with abnormal pulmonary status.
• These study animals were not treated at the feedlot, they might have potentially resolved any pulmonary lesions existing at the time of clinical (visual) scoring, or, were false positive designations at the time of clinical illness scoring.

Benefits to industry

This project has evaluated the accuracy of a range of BRD screening and confirmatory diagnostics to detect lung pathology in live animals. Ultimately if accurate BRD diagnostics can be identified, early treatment and recovery will be achieved, maximising health, welfare, and productivity. The correct animals will be treated benefiting antimicrobial stewardship.

This project was the first detailed evaluation of thoracic ultrasound for the Australian feedlot industry. Ultrasound performs well as a BRD diagnostic method but requires training to conduct the surveys and interpret images, minimum machine capability, an appropriate restraint facility that allows access to both sides of the chest and the forward chest area, and a minimum of five minutes available to conduct the more basic level survey.

It should also be noted that presumptive CASEs presented for examination had demonstrated a minimum of three clinical illness scores of 2 or greater over a 24-hour period. These CASEs therefore represent moderate to severe morbidity where the disease (and lung lesions) are allowed to progress unlike in typical feedlot clinical settings. Nonetheless, this diagnostic modality showed the best accuracy.

For commercial application, where a truncated survey would be required to save time, prioritising anteroventral (forward and low) positions on the right hemithorax (right chest side) would yield the best probability of detecting pulmonary lesions if they are present.

MLA action

The results of this project will inform MLA's future research in automation of Bovine respiratory disease detection. Results will be presented to the ALFA/MLA Consulting Veterinarian and Nutritionist meeting.

Future research

Future research will focus on practical utility and these diagnostics to diagnose stage and severity of lung pathology over larger numbers of feedlot cattle. Ultimately if accurate confirmatory diagnostics can be identified, these can be utilised to calibrate screening diagnostics algorithms leading to automation of disease detection.