Estimation of the accuracy of two diagnostic methods for the detection of Plum pox virus in nursery blocks by latent class models

摘要

The control of Plum pox virus (PPV), the most important viral disease that affects stone fruit trees, requires the use of reliable detection methods. The effectiveness of spot real-time reverse transcriptase polymerase chain reaction (RT-PCR) for the detection of PPV in samples collected from nursery blocks was compared with a validated PPV detection technique, the double antibody sandwich indirect enzyme-linked immunosorbent assay (DASI-ELISA) using the PPV-specific monoclonal antibody 5B-IVIA/AMR. In total, 5047 nursery plants were analysed by both techniques. The agreement between the techniques was almost perfect (Cohen's kappa index of 0·88 ± 0·01). The diagnostic parameters (sensitivity, specificity and likelihood ratios) of both techniques were simultaneously evaluated in 2473 nursery plants by latent class models using maximum likelihood functions and a Bayesian approach. The sensitivity and specificity of both techniques did not vary according to the latent model applied. Spot real-time RT-PGR was more sensitive while DASI-ELISA was more specific for PPV detection. In addition, the findings demonstrate that latent class models are a flexible and potent statistical method to estimate the accuracy of diagnostic tests for plant pathology.