RESUMEN
Although avian influenza (AI) virus of H5 and H7 subtypes has the potential to mutate to a highly pathogenic form and cause very high mortalities in some poultry species, most AI infections in poultry are due to low pathogenic AI (LPAI). Hence serological surveys, coupled with passive surveillance activities, are essential to detect sub-clinical infections by LPAI viruses, H5 and H7 subtypes. However the proper planning of an active surveillance system should be based on a careful estimation of its performance. Therefore, the sensitivity of the active surveillance system for AI in the western region of Cuba was assessed by a stochastic model quantifying the probability of revealing at least one animal infected by H5 or H7 subtype. The diagnostic sensitivity of the haemagglutination inhibition assay and different levels of within-flock prevalence (5%, 12% and 30%) were considered. The sensitivity of the surveillance system was then assessed under five different samples size scenarios: testing 20, 30, 40, 50 or 60 animals in each flock. Poultry flock sites in the western region of Cuba with a size ranging from 10,000 to 335,000 birds were included in the study.