Evaluation of serological tests for H5N1 avian influenza on field samples from domestic poultry populations in Vietnam: consequences for surveillance.

In Vietnam, serological post H5N1 vaccination surveillance using the HI test is applied to assess the efficiency of the vaccination in addition to virological monitoring. In this paper we report on the evaluations of the performances of the haemagglutination inhibition (HI) test and of a H5-ELISA, using chicken and duck field samples. The evaluations were conducted by comparison with a pseudotyped-based virus neutralization test (H5pp VNT) performed in a reference laboratory and considered as a "gold standard" and also by using methods developed for imperfect reference test. Their global accuracy and best cut-offs were also estimated. Results from the HI test for several haemagglutinin subtypes and from a commercial type A influenza competition ELISA were also compared. The results showed that performance of the HI test was very good in comparison with the H5pp VNT. Data also clearly supported the cut-off of ≥ 4 log(2) used for the HI test for chickens but, a 3 log(2) positivity cut-off would be more appropriate for ducks. When compared with the VNT, the H5-ELISA showed poor specificity when using the positivity cut-off specified by the manufacturer but could be used as a screening test if confirmed by the HI test or the H5ppVNT which presents some interests for large scale testing (no need for biosafety level 3 conditions and high performance). A general and highly sensitive pre-screening can also be achieved using the detection of NP-specific antibodies with a competition ELISA. This appears of little interest in a context of high subtypes diversity where only a subtype is targeted for surveillance and control.