Avian Influenza A (H5N1) Virus and 2 Fundamental Questions

Since 1997, when the avian influenza A (H5N1) virus infected 18 residents of Hong Kong, Special Administrative Region, China, 6 of whom died, 2 questions have been of paramount importance related to this virus. How many other individuals are infected? How are individuals becoming infected?

Since the first cases, some parts of the answers to these questions have been evident. When the first index case occurred in Hong Kong in May 1997 and was followed by 17 cases in November and December, there were preceding and contemporary outbreaks of H5N1 virus infection among poultry in the area. Viruses isolated from the infected persons were virtually identical to those obtained from infected poultry, and all of these viruses appeared to be avian in origin on the basis of genetic sequencing. Largely on the basis of that limited information, in addition to the insights gathered from field investigations, public health and agricultural authorities took several decisive and effective steps. These included culling of poultry, enactment of measures to limit virus propagation in the wet markets, and communications to the public. In combination, these measures rapidly and dramatically stopped the immediate outbreak both among poultry and humans. Within days after the last human cases, formal epidemiological field investigations were started. The findings from those studies suggested that few contacts of infected individuals had evidence of infection [1] and that exposure to wet markets and, by logical extrapolation, to infected poultry or a contaminated environment was a major identifiable risk factor for zoonotic H5N1 virus infection [2].

In 2003, H5N1 virus infections in humans were again identified in Hong Kong—this time, in a father and son who had recently returned from a visit to Fujian Province in southern China. A daughter had died of an undiagnosed febrile illness during the same trip. Although these cases raised concerns among scientists looking for signs of the next influenza pandemic, they were soon overshadowed by the emergence of a new and unrelated disease, severe acute respiratory syndrome.

In the beginning of 2004, after an extraordinary global effort had ended the spread of severe acute respiratory syndrome, several countries in Southeast Asia reported, in close succession, outbreaks of H5N1 among poultry. Human infections were reported in Vietnam and Thailand. In retrospect, these events heralded a major surge in the spread of H5N1 viruses that began in 2003. For example, by the end of 2005, the virus had been detected in poultry and humans in Europe (Turkey); by March 2006, the virus had been detected in poultry and humans in Africa (Nigeria); and from the end of 2007 through early 2008, human cases were reported from Pakistan, Myanmar, and Bangladesh [3].

Despite the relatively limited numbers of human infections, H5N1 viruses have been widely recognized as major global public health and agricultural concerns. The reasons are clear. As a pathogen, H5N1 has posed a triple threat to a degree significantly beyond that of any previously known influenza virus. Among animals, this highly pathogenic avian influenza virus has caused the death of millions of poultry. As such, it has threatened the agricultural sectors of some countries and the food supply of some communities. As a zoonotic influenza infection, H5N1 virus infection has been uniquely lethal. Since 2003, >400 people have had laboratory‐confirmed disease, with an overall case‐fatality rate of 63%. Although the most recent case‐fatality estimates range from 42% in Egypt to 82% in Indonesia and the estimates have varied over time, they have remained remarkably consistent and high. However, the most feared aspect of H5N1 is the possibility that this virus, which continues to evolve, might develop the capacity to sustain human‐to‐human transmission and, thereby, spread worldwide. Influenza pandemics have recurred over centuries, and a future pandemic is anticipated. The potential to cause a pandemic is not unique to H5N1 but probably is inherent in many animal influenza viruses. For example, avian H7 and H9 influenza viruses have also caused zoonotic infections. In 2003, an outbreak of H7N7 infection among poultry in The Netherlands resulted in 89 zoonotic cases and 1 death. Potential pandemic viruses might include swine influenza viruses, as well as other animal influenza viruses.

When placed in the modern global context in which societies are highly interconnected by the rapid movement of people, animals, and goods and by the even more rapid transmission of information through media and the Internet, H5N1 and its associated agricultural, zoonotic, and pandemic challenges exemplify the complexities posed by global emerging infectious diseases. As such, the need to understand H5N1 is more compelling today than it was in 1997.

In some respects, the broad epidemiology of H5N1 virus infection has become clearer. Since 2003, infections in either poultry or wild birds have been reported in 61 countries. Human infections have generally paralleled the incidence of outbreaks among poultry and have demonstrated a seasonal pattern. Similar to seasonal influenza viruses, H5N1 has infected male and female individuals in roughly equal numbers, although patterns differ by country. Annual numbers of human cases reported to the World Health Organization have decreased from 109 in 2006 to 88 in 2007 to 44 in 2008. The decrease is encouraging, because it probably reflects better control of the virus among poultry populations, as well as greater awareness and, hopefully, better avoidance of unnecessary exposure to infected birds. However, the occurrences of recent cases in 2009 in countries such as China and Vietnam, where good poultry control programs have been established, should indicate that this virus is still very active and highly persistent. Moreover, it would be wrong to construe the decrease in the number of human cases as evidence that the risk of a pandemic posed by this virus has decreased. It is simply unknown whether the decrease in the number of officially reported human cases meaningfully reflects any change in the potential for this virus to evolve into a pandemic virus.

In addition, much information about the epidemiology of this virus remains unknown. For example, 15 countries have reported human infections, but the majority of cases (90%) have occurred in 5: Indonesia, Vietnam, Egypt, China, and Thailand. It is uncertain why human cases have been concentrated in these countries and few or no human cases have been reported in other countries where the virus has been repeatedly identified among poultry populations. Moreover, within countries, the distribution of cases has been markedly nonuniform. This might be related primarily to differences in the distribution of infected birds but probably also reflects important differences in additional factors (e.g., levels of surveillance and local behaviors). A related question is: why does the number of human cases appear relatively limited in comparison with the presumably high number of exposures to infected poultry? In part, the answer is that H5N1 remains an avian influenza virus that is not adapted to humans. However, many other contributing factors are possible. One particularly important hypothesis is that cases are being missed because current surveillance primarily detects severe infections. It is clear that the reported numbers of laboratory‐confirmed human cases are conservative, but the extent to which these numbers are conservative is unknown.

The current issue of the Journal contains 3 articles that address these issues. In the study conducted in Cambodia by Vong et al. [4], a serological survey was conducted among 674 villagers living in close proximity to 2 persons who died of H5N1 virus infection during 2006. The survey found 7 previously unidentified persons (1% of the total) with high antibody titers to H5N1. The authors concluded that the overall prevalence appeared low, generally in agreement with other serological surveys. This conclusion is somewhat reassuring, because it suggests that large pockets of infected persons are probably not being missed. However, the risk of infection also remains difficult to estimate, because the knowledge of what constitutes a meaningful exposure to the pathogen remains limited. What, in other words, should be the denominator for estimating the risk of infection? Moreover, 7 (78%) of 9 infected persons in the 2 Cambodian villages were missed by current surveillance. Although these results cannot be considered to be representative without broader confirmation, they show that, in some settings, surveillance may substantially miss H5N1 virus infections. On balance, identifying more human H5N1 virus infections probably would not change most current measures or recommendations to prevent zoonotic H5N1 virus infection. By contrast, significantly undercounting cases could be an important limitation for the earliest possible detection of an H5N1 pandemic. In the associated nested case‐control study, H5N1‐seropositive individuals were more likely to report bathing or swimming in contaminated ponds near free‐grazing ducks than were H5N1‐seronegative individuals. However, the interpretation of these findings was severely limited by the long delay between the periods of interest and the interviews.

In the second study, Zhou et al. [5] conducted a case‐control investigation in China for risk factors for H5N1 virus infection among 28 persons. The multivariate analyses identified direct contact with sick and dead poultry, visits to live poultry markets, and indirect contact (defined as being within 1 m of poultry, poultry products, or poultry feces but without physical contact) as risk factors. These findings support earlier studies and underscore the current understanding that exposure to infected poultry or virus‐contaminated environments are risk factors for zoonotic infection. The authors also highlighted the important role, especially in an urban setting, of wet markets. In some H5N1 virus–affected countries, wet markets or live animal markets appear to play a pivotal role in the dynamics of this virus by facilitating cross‐species infection, virus amplification, and exposure of people to infected birds and contaminated environments. The primary limitation of this study was the time lag of >2 years in some instances between the administration of the questionnaire and the study period of interest.

In the third study, performed in Thailand by Tiensin et al. [6], the investigators looked for ecological risk factors associated with spatial and temporal clusters of H5N1 outbreaks among poultry. They found that subdistricts with high flock density of fighting cocks, quail and free‐grazing duck flocks, and poultry slaughterhouses had higher infection rates. The study did not include live bird markets because they are uncommon in Thailand; however, the authors also concluded that poultry slaughterhouses, similar to live poultry markets, functioned as dissemination points for infection and could be targeted for preventive measures. Although this study did not try to identify risk factors for individual zoonotic infections, the findings are relevant for public health efforts to prevent such infections.

Together, the 3 studies do not change overall concepts of how frequently individuals are infected or the idea that human infections are directly or indirectly linked with poultry infections. However, within that overall framework, these studies provide important details and underscore the notion that the dynamics and risk factors for this infection vary depending on the setting. One take‐home lesson is that, to be effective, public health guidance should be tailored to the local situation. A second lesson is that we still do not understand enough about the specific details of how individuals are infected. For example, case level studies conducted much closer to the time of infection are needed if exposures are to be explored in more detail. The need for this level of understanding is more compelling now than it has ever been.

• Received February 18, 2009.
• Accepted February 18, 2009.

• © 2009 by the Infectious Diseases Society of America. All rights reserved.