Comparative evaluation of four competitive/blocking ELISAs for the detection of influenza A antibodies in horses.

New Zealand is free from equine influenza and has never experienced an incursion in its horse population. As part of New Zealand's preparedness to an incursion of an exotic animal disease, it was considered necessary to select the most accurate test for equine influenza (EI) from the array of those available. Four readily available blocking/competitive enzyme-linked immunosorbent assays (ELISA), originally developed and marketed for the detection of antibodies against the avian influenza virus, were evaluated using serum samples from New Zealand non-infected, non-vaccinated horses (n=365), and Australian field infected (n=99) and experimentally infected horses (n=3). Diagnostic specificities (DSP) and diagnostic sensitivities (DSE) were determined as follows: ELISA-1=98.1%/99.0%; ELISA-2=90.1%/99.0%; ELISA-3=98.1%/96.0%; ELISA-4=95.3%/99.0%. For ELISA-1, DSP and DSE results were comparable to previously published data on a larger sample number from Australian horses (Sergeant et al., 2009). Receiver operating characteristics (ROC) and frequency histogram analysis were also performed. The area under the curve (AUC) ranged from 0.996 to 0.979, with ELISA-1 possessing the highest AUC, followed by ELISA-2, ELISA-4 and ELISA-3. Separation of the negative and the positive serum panel was best for ELISA-4, followed by ELISA-2, ELISA-1 and ELISA-3. In three experimentally infected horses, sero-positivity was detected between 7 and 9 days post-infection, with ELISA-4 being most sensitive, followed by ELISA-1, ELISA-2 and ELISA-3. Overall, the four ELISAs performed well in this evaluation but some differences were observed.