Diagnostic performance of direct and indirect methods for assessing failure of transfer of passive immunity in dairy calves using latent class analysis.

Accurate diagnosis of failure of transfer of passive immunity (FTPI) in newborn calves is an essential component of dairy farm management plan. Several methods (direct and indirect) are available for diagnosis of FTPI in dairy calves. However, the indirect methods offer an advantage over the direct methods in not requiring an experienced veterinarian, rapid, cost efficient and can be performed under field-setting. The objective of this study was to estimate the diagnostic performance of radial immunodiffusion (RID) assay, transmission infrared (TIR) spectroscopy and digital Brix refractometer for diagnosis of FTPI in dairy calves using latent class models at four cut-off values of digital Brix refractometer. Holstein calves (n = 691) from 40 commercial dairy farms in the four Atlantic Canada provinces were blood-sampled and tested for detection of FTPI. Results showed that the number of calves with FTPI was 253 (36.6%) by RID, 194 (28.1%) by TIR and 204 (29.5%) by Brix refractometer at cut-off value of 8.2%. Estimates of Se was higher than Se and Se, at all Brix refractometer cut-offs, but with increase of Brix refractometer cut-off from 8.2 to 8.5%, Se and Se were decreased from 96.0% (95% PCI: 88.0-99.0) and 79.0% (95% PCI: 70.0-85.0), to 92.0% (95% PCI: 77.0-99.0) and 74.0% (95% PCI: 61.0-82.0), respectively. Sp and Sp were always higher than Sp at all tested cut-offs and were above 92.0%, and 96.0%, respectively. With increasing the cut-off of Brix refractometer from 8.2 to 8.5%, Se estimate has remarkably increased from 79.0% (95% PCI: 70.0-96.0) to 95.0% (95% PCI: 87.0-100.0), respectively. Whilst, Sp was decreased from 95.0% (95% PCI: 91.0-98.0) at cut-off 8.2% to 84.0% (95% PCI: 78.0-94.0) at cut-off 8.5%. In conclusion, RID has a higher Se than TIR and Brix, if the latter is used with cut-offs of 8.2% or 8.3%. However, the higher the cut-off, the more comparable sensitivities of RID and digital Brix refractometer. The median estimate of Sp was always higher than Sp and Sp at all tested cut-offs. However, the 95% confidence interval estimates of the three tests were overlapping across the tested cut-offs of digital Brix refractometer reflecting the inability to prefer a test over the other based on the Sp estimate.