From December 1999 through April 2003, >50 million birds died or were depopulated after HPAI infection in the European Union (1), causing severe economic losses to the private and public sectors. These losses suggest that the strategies and control measures used to combat the disease need improvement, from disease control and animal welfare perspectives.

AI viruses are introduced to domestic poultry primarily through direct or indirect contact with infected birds. Transmission may occur through movement of infected poultry; movement of contaminated equipment, fomites, or vehicles; and exposure to contaminated infectious organic material. Airborne transmission over long distances between farms has not yet been demonstrated. For these reasons, if biosecurity measures are implemented at the farm level, AI infections can be prevented.

Outbreaks that involve large numbers of animals are characterized by the penetration of infection into the commercial circuit; that is, industrially reared poultry and all other poultry that is traded, including those from semi-intensive and backyard farms. Biosecurity (encompassing bioexclusion and biocontainment) represents the first and most important means of prevention. If biosecurity measures of a high standard are implemented and maintained, they create a firewall against infection penetration and perpetuation in the industrial circuit. However, breaches in biosecurity systems do occur. On one hand, the occurrence and extent of the breach should be evaluated and corrective measures should follow; on the other, they indicate the need to establish early warning systems and additional control tools for AI.

Until recently, AI infections caused by viruses of the H5 and H7 subtype occurred rarely, and vaccination was not considered because stamping out was the recommended control option. Primarily for this reason, vaccinology for AI has not grown at the same rate as for other infectious diseases of animals. Data are being generated from experimental and field research in AI vaccinology, but the rather complex task of vaccinating poultry in different farming and ecologic environments still has areas of uncertainty.

Guidelines on disease prevention and control have been issued as joint recommendations of the World Organization for Animal Health (OIE), the Food and Agriculture Organization (FAO), and the World Health Organization (8). These recommendations, however, need to be put into practice in a variety of different field situations; the applicability of 1 system rather than another in a given situation must be evaluated, weighing the benefits of a successful result against the drawbacks of failure.

Vaccination can be a powerful tool to support eradication programs if used in conjunction with other control methods. Vaccination has been shown to increase resistance to field challenge, reduce shedding levels in vaccinated birds, and reduce transmission (9,10). All these effects of vaccination contribute to controlling AI; however, experience has shown that, to be successful in controlling and ultimately in eradicating the infection, vaccination programs must be part of a wider control strategy that includes biosecurity and monitoring the evolution of infection.

To eradicate AI, the vaccination system must allow the detection of field exposure in a vaccinated flock, which can be achieved by using conventional inactivated vaccines and recombinant vector vaccines. Conventional inactivated vaccines that contain the same viral subtype as the field virus enable detection of field exposure when unvaccinated sentinels left in the flock are tested regularly. This system is applicable in the field but is rather impracticable, especially for the identification of sentinel birds in premises that contain floor-raised birds. A more encouraging system, based on the detection of anti-NS1 antibodies, has been recently developed and can be used with all inactivated vaccines, provided they have the same hemagglutinin subtype as the field virus (11). This system is based on the fact that the NS1 protein is synthesized only during active viral replication and, therefore, is rarely present in inactivated vaccines. Birds vaccinated with such vaccines will develop antibodies to NS1 only after field exposure. Full and field testing of this system under different circumstances are still in progress (11,12), and results should be available before this system is recommended.

To date, the only system that enables detection of field exposure in a vaccinated population and that has resulted in eradication is based on heterologous vaccination and known as "DIVA" (differentiating infected from vaccinated animals). This system was developed to support the eradication programs in the presence of several introductions of LPAI viruses of the H7 subtype (1,9). Briefly, a vaccine is used that contains a virus possessing the same hemagglutinin, but a different neuraminidase, as the field virus. This vaccination strategy enables detection of antibodies to the neuraminidase antigen of the field virus. For example, a vaccine containing an H7N3 virus can be used against a field virus of the H7N1 subtype. Antibodies to H7 are cross-protective, thus ensuring clinical protection, increased resistance to challenge, and reduction of shedding, while antibodies to the neuraminidase of the field virus (in this case N1) can be used as a natural marker of infection. Experimental data on the quantification of the vaccination effect on transmission within a flock indicate that the reproduction ratio can be reduced to <1 by 1 week after vaccination (10). Such a reproduction ratio indicates minor rather than major spread of infection. In simple terms, such vaccination interventions will substantially reduce (although not prevent) secondary outbreaks, depending on the immune status of contact birds and flock.

Promising results have also been obtained with vaccines generated by reverse genetics (13). These vaccines are expected to perform like conventional inactivated vaccines; however, data are not yet available as to their efficacy under field conditions. Recombinant fowlpox vaccines that express the hemagglutinin protein of the field virus have also been reported to be efficacious for reducing shedding levels and providing clinical protection (14). They enable the detection of field exposure because vaccinated unexposed animals do not have antibodies to any of the other viral proteins. Any test developed to detect antibodies to the nucleoprotein, matrix, NS1, or neuraminidase of the field virus can be used to identify field-exposed birds in a vaccinated population. However, the performance of these vaccines in relation to the immune status of the host to the vector virus is unclear (15). Recent encouraging studies indicate that vaccination of day-old chicks with maternal antibodies against fowlpox has been successful. Data are lacking on the performances of such vaccines in a population that has been field exposed to fowlpox. Another aspect that must be carefully considered is the host. These vaccines are likely to induce protective immunity only in birds that are susceptible to infection with the vector virus.

Regardless of the vaccine and companion test used, mapping occurrence of infection within the vaccinated population is imperative, primarily to monitor the evolution of infection and to appropriately manage field-exposed flocks. Field exposure represents a means by which infectious virus may continue to circulate in the immune population; for this reason, vaccination can be considered as only part of a control strategy based on biosecurity, monitoring, approved marketing procedures, and stamping out. An inappropriately managed vaccination campaign will likely result in the virus becoming endemic.

Inadequate biosecurity or vaccination practices can lead to transmission between flocks and selection of variants that exhibit antigenic drift. Antigenic drift of H5N2 viruses belonging to the Mexico lineage, resulting in lower identity (less similarity) to the vaccine strain, has been described (16). Extensive use of vaccine in Mexico has resulted in the emergence of antigenic variants that escape the immune response induced by the vaccine. This occurrence is similar to antigenic drift that typically occurs in animals with a long lifespan (pigs and horses) that are routinely vaccinated and in human beings. Mexico has been vaccinating poultry since the HPAI outbreak in 1994 without applying the DIVA principle. Although no HPAI virus has been reported since the implementation of the vaccination campaign, LPAI viruses continue to circulate. Conversely, a similar approach in Pakistan after the HPAI H7N3 outbreaks in 1995 resulted in the isolation of HPAI H7N3 virus ≈10 years later, in 2004 (17).

The international scientific community is debating how vaccination of poultry would affect human health. On one hand, vaccinated birds shed less virus; on the other, they do not show any clinical signs of disease and could therefore act as silent carriers. Several factors contribute to the development of infection in humans: insufficient hygienic standards, the characteristics of the strain, and presence of viral dose sufficient to infect a human being. The possibility that vaccinated poultry may not shed enough virus to infect a human being is substantiated by recent field evidence. With reference to the H5N1 crisis, several countries are using vaccination to support control efforts. Vietnam implemented a nationwide vaccination campaign, which was completed in early 2006. The campaign's main achievement is that despite 61 cases of human infection between January and November 2005, no human cases of AI have been reported in Vietnam after December 2005 (18).

The scientific veterinary community must control AI infections in poultry for several reasons: to manage the pandemic potential, to preserve profitability of the poultry industry, and to guarantee food security to developing countries. Although biosecurity is recognized as an excellent means of preventing infection, in certain situations the biosecurity standards necessary to prevent infection are difficult to sustain. Vaccination is a potentially powerful tool for supporting eradication programs by increasing the resistance of birds to field challenge and by reducing the amount and duration of virus shed in the environment. Vaccination strategies that encompass monitoring of infection in the field are crucial to the success of such efforts.

Timely information is needed about the efficacy of vaccination in a variety of different avian species, bearing in mind the diverse farming systems used in developed and developing countries. The outcome of such efforts should be made available to the international community because decision makers lack enough information to make educated choices. An enormous effort is required from national governments and funding bodies to make resources available to research programs to develop improved control measures that can be applied under different local conditions. To maximize the global effort to combat this disease, developing and sustaining transversal research programs on AI control, which encompass veterinary and agricultural science, are imperative.