Evaluation of nasal swab and nasal wipe for detection of Influenza A in swine using Bayesian latent class analysis

Abstract

Influenza A virus (IAV) is an important pathogen in Brazilian swine herds, and monitoring the viral circulation is essential to control and reduce the transmission. Surveillance programs for IAV are often based on individual piglets level sampling, making the evaluation of the available diagnostic tools crucial to assessing IAV circulation in herds. Thus, two sample collection methodologies were compared in pig herds in southern Brazil to detect IAV by RT-qPCR: nasal swab (NS) and nasal wipe (NW). A Bayesian latent class model (BLCM) was set for two tests and two populations. The NW and NS used are more specific (higher than 95 % for both) than sensitive. The sensitivity for NW was lower than the NS, 84.14 % (70 % – 95 %; posterior probability interval (PPI): 95 %) and 87.15 % (73 % – 97 %; PPI: 95 %), respectively, and the specificity was 95 % (90 % – 99 %; PPI: 95 %) and 99 % (96 % – 100 %; PPI: 95 %), respectively. Although the wipe sample collection loses both sensitivity and specificity compared with nasal swab, differences in test performance were very limited and PPIs largely overlapped. Therefore NW can also be considered a valuable tool. The decision about the use of both techniques should be based on the trade-off between their performance limitations and feasibility in routine monitoring.