Harvard Public Health Review
Avian Flu: Preparing for a PandemicBy Marc Lipsitch, associate professor of epidemiology, and Barry R. Bloom, dean of HSPH
The infectious disease community has been concerned about the possibility of pandemic influenza for almost two decades, and the U.S. government began to draft a Pandemic Emergency Plan in 1991. But it was not until the aftermath of Hurricane Katrina that our country's lack of preparedness for handling major disasters became evident to all. If any good can come of this season's hurricanes, the tsunami in Indonesia, and the earthquake in Pakistan, it is a recognition of the vulnerability of all countries to natural disasters--including pandemic infectious diseases--and the need to plan in advance.
The most obvious threat at the moment is the avian influenza virus known as A/H5N1. This pathogenic subtype has caused flu outbreaks in birds in at least 15 countries, from Croatia to Japan and Indonesia, killing or causing the culling of over 140 million birds. Alongside these massive avian outbreaks, there have been 134 confirmed human cases, more than half of whom have died. Nearly all of these human cases are traceable to exposures to infected birds; mercifully, H5N1 thus far lacks the ability to be transmitted efficiently between humans. Spurred by the concern that this or another strain of influenza virus could spark a pandemic in the coming years, President George W. Bush outlined on November 1 his $7.1 billion plan, almost 50 times the current funding level, for pandemic preparation. Recognizing that no single measure can stop a pandemic or blunt its impact, the plan appropriately includes funding for a wide range of activities. The threat of a flu pandemic has policymakers' attention. Now it is essential that the new resources be used wisely to deal with the threat at all levels, from local public health to international surveillance.
Scientists cannot predict whether or when a mutation might occur in the H5N1 bird flu strain--or in any of 20 other avian influenza strains--that would enable it to be transmitted readily to and between humans. What would a pandemic look like if H5N1, or another strain, were to acquire the capacity for extensive human-to-human transmission while remaining highly virulent? The best model we have for such a scenario is the 1918 influenza pandemic. That year at Camp Devens, Mass., there were 12,604 cases and 727 deaths within just two weeks of the first case, in September. Within five months, a quarter of the U.S. population was struck and about one in 200 Americans died. Worldwide, 20 million to 50 million people succumbed to the disease.
Since 1700, there have been influenza pandemics that have spread rapidly and widely at irregular intervals at least three times a century, the last pandemic being in 1968, 37 years ago. In one of the great scientific detective stories of all time, scientists at the U.S. Centers of Disease Control and Prevention (CDC) obtained the genome sequences of 1918 influenza from samples isolated from a person who perished and was preserved in the Alaskan permafrost, and from pathology specimens saved at the Armed Forces Institute of Pathology since 1918. Studies of a virus reconstructed from these genome sequences have recently revealed many features that contributed to the virus's unusual virulence. Another inference from the molecular genetic analysis is that the 1918 flu was very likely a bird flu.
The last emerging virus that threatened to spark a potential pandemic was SARS, which a team of HSPH epidemiologists, including Megan Murray, James Robins and one of the present authors, Marc Lipsitch, modeled in real time. Their model of transmission indicated that the virus's transmissibility (R0 value) was such that each infectious case gave rise to about 3 secondary infectious cases, while the average time between the primary and secondary cases (serial interval) was about 8.4 days. If appropriate isolation of infectious patients and quarantine of exposed contacts were carried out, the model predicted that transmission would be broken. As the researchers were drafting their paper, the evidence in Singapore (and, later, mainland China, Hong Kong, Toronto, and elsewhere) proved that combining those interventions was working to diminish transmission--and the epidemic was controlled.
Unfortunately, the same models predict that a bird flu capable of efficient transmission between humans is likely to be much more difficult to control than SARS. Analyses of the 1918 flu pandemic by an HSPH graduate student, Christina Mills, working with our infectious disease epidemiology team, revealed that the 1918 flu virus was no more, and possibly less, transmissible between humans than the SARS virus (R0 = 2-3). But the time between infectious cases, or the serial interval, as measured in seasonal flu, is estimated to be approximately four days, with people being infectious as early as the first or second day after infection, often before they are aware they have the flu. Notably, even in the fatal cases of H5N1 observed in Asia so far, the earliest symptoms were also relatively mild, only becoming life-threatening after a week or longer.
These contrasts in the natural history of flu and SARS have striking implications for the current planning for a possible pandemic. With so little time for the public health system to intervene effectively, the tools that worked so well for SARS--isolation of infected individuals, and quarantine of persons exposed to infected individuals--are simply unlikely to be very effective in the case of a virulent flu pandemic. In the case of SARS, screening at international borders proved to be of limited benefit because so few infected people were traveling (in all, SARS infected just over 8,000 people). For flu, the rapid spread characteristic of a pandemic will cause just the opposite problem for border controls: many infected people will likely be traveling, with many not yet showing signs of infection. With over a million travelers entering the United States each day, stopping 95 or even 99 percent of infected people at borders would not be sufficient once a pandemic was under way in other parts of the world. At least a few infected individuals would pass undetected through even the best-organized system, carrying the potential to seed new epidemics. Governments will likely feel strong political pressure to try to localize a pandemic and keep it outside their borders, but such efforts are likely to fail rapidly. Once an epidemic is under way in a particular place, further efforts to stop importation will be of little benefit.
How can the world protect itself against the next pandemic? There are drugs that, if taken within two days of the onset of symptoms, will reduce the duration and severity of illness, based on our understanding of how they affect current seasonal flu strains. Although these drugs work against H5N1 in cell culture, their clinical value in H5N1 patients is unclear, and mutants with some resistance to the major drug, oseltamavir (Tamiflu), have already emerged. More to the point, existing supplies of this drug, even in the richest countries, will be orders of magnitude too small to curb transmission of H5N1 once a pandemic is under way. A few countries, like the UK, have ordered enough Tamiflu to treat up to a quarter of the population and perhaps reduce the morbidity and mortality associated with the infection. But in most wealthy countries, including the U.S., and in developing countries, supplies are far more limited. The World Health Organization (WHO) is planning to try to snuff out an incipient pandemic while outbreaks remain localized, but even the strongest proponents of such a strategy admit that the challenges are formidable.
Other measures are urgently needed to reduce the number of introductions of H5N1 viruses into humans, thereby reducing the risk that a pandemic-ready strain could emerge as the virus adapts genetically to people. Despite the massive culling of infected birds to date, international health and agriculture authorities have stated that current funding is inadequate to give farmers an incentive to report infections in poultry. These authorities also indicate that farmers, veterinarians, and health care workers in rural areas need basic education about how to identify potential animal and human cases, and how to minimize the risk that the infection will be transmitted. China's Minister of Health, Gao Qiang, was asked during his visit to HSPH in October whether China was covering up cases of avian flu. He denied such a cover-up, but stated that he was seriously concerned that cases might be going unreported, given that local officials do not know what to look for or might not report cases quickly enough to the central government.
Surveillance efforts in Asia are critical to reducing the number of human cases and delaying the emergence of a pandemic as we undertake preparations at home. Such efforts are also essential to give us early warning of changes in the drug-resistance patterns and the immunological determinants of circulating strains--information that will permit drug and vaccine strategies to keep up with the evolution of the virus. Despite the clear need for such efforts, less than 5 percent of the budget in President Bush's plan is directed to overseas surveillance.
If a pandemic does come--as most scientists believe it will, despite our best efforts--there will be an urgent need to support responses at the community level. Like politics, all infectious diseases start locally. As generous as the President's proposed plan may be for developing drugs and vaccines against pandemic infectious diseases, primarily influenza, it has only a small portion, $500 million, for strengthening public health responsiveness in the 50 states. Howard Koh, former commissioner of Public Health for the Commonwealth of Massachusetts and head of our Division of Public Health Practice, with support from the CDC, has created a Center for Public Health Preparedness. David Gergen of Harvard's Kennedy School of Government, working with the HSPH Center's Leonard Marcus, has launched a program for national leadership training known as the National Preparedness Leadership Initiative. These programs are providing the best information we have at Harvard University for equipping government officials with the leadership skills they need to deal with emergencies and pandemics.
According to official Department of Health and Human Services projections, were a pandemic to occur, 200 million Americans could be infected, and perhaps a half-million would require hospitalization. If isolation and quarantine are unlikely to be very effective, we will require a host of alternative protection strategies, ranging from emphasizing hand washing, providing N95 respirators, and reducing social proximity, to preparing workplaces to keep employees at home, protecting medical personnel and first responders, providing treatments, and expanding hospital and mortuary surge capacity in cities and states.
As we undertake measures to reduce the risk of a pandemic and bolster our ability to respond if one does occur, developing an effective vaccine and the capacity to manufacture it in sufficient quantities must form the centerpiece of our preparations. Other public health interventions can postpone transmission of the virus or mitigate its effects, but vaccines are unique in their ability to prevent and provide long-term protection against infection with a new strain of influenza. As Anthony Fauci, the director of the National Institute of Allergy and Infectious Diseases (NIAID) put it, "The biggest challenge unequivocally is vaccine production capacity." The U.S. can produce about 60 million doses of vaccine for seasonal flu each year; the world can produce perhaps 300 million. Yet the H5N1 vaccine recently tested by NIAID required at least four times as much killed virus to produce a satisfactory immune response as is normally used in vaccines. That means we could expect to make only 15 million doses in the U.S. and 75 million worldwide--far short of the need in a pandemic.
How can we stretch this capacity? The single largest item in the president's proposal is $2.8 billion to develop a vaccine-production system that is cell-culture-based. Today, we make flu vaccines in chicken eggs, using an approach that dates back half a century and depends on adequate egg supplies and specialized equipment. Developing a new technology to grow the vaccine seed strain in cell culture, as other viral vaccine strains are grown, would be a major advance--one that would speed vaccine production and expand capacity by allowing emergency conversions of vaccine-production facilities from other viruses to influenza in the event of a pandemic. However, this strategy will have a payoff only in the medium- to long-term, at least five years from now.
In the short run, one promising approach that could expand our manufacturing capacity is the use of adjuvant-containing vaccines. Adjuvants are chemical additives that augment the immune response, allowing smaller doses of vaccine to be effective and the existing production capacity to yield more doses. An exciting pilot study of an adjuvanted vaccine against influenza H5N3, a relative of the current avian strain, showed that two doses together containing one-third as much antigen as a normal flu vaccine, plus a novel adjuvant manufactured by one vaccine company, could produce a strong immune response. Equally important, this immune response appeared to be broadly effective across a range of different H5 viruses, suggesting that the adjuvant could make a vaccine effective even if it were not perfectly matched against the precise emergent pandemic strain. Alum, an aluminum salt that has been used as an adjuvant since the early 20th century, also appears to be an effective adjuvant in flu vaccines, and has the advantage of having been licensed for many years for use in the U.S. Clinical trials of adjuvanted H5N1 vaccines are now under way, but additional trials and regulatory hurdles must be cleared before such vaccines can be made ready for general use. If accelerated, this process could provide capacity to manufacture flu vaccine for nearly the entire U.S., and could also stretch capacity to provide doses for a significant proportion of the world. Careful clinical studies of vaccines with adjuvants will be essential, however, since an adjuvant that produces too strong an immune response to the infecting virus or an inappropriate immune response could conceivably increase pathology and exacerbate disease.
If we had a vaccine, how could we get people to take it, given that less than a quarter of the U.S. population receives seasonal flu vaccine in a given year? Indeed, even with a shortage, more unused seasonal flu vaccine was discarded in 2005 than in 2004. One strategy for introducing a bird flu vaccine, once we had a sufficient supply, would be to add it as a component of our trivalent seasonal influenza vaccine. We do better against seasonal flu, immunologically speaking, in part because we have already developed some immunity and immunological memory to related strains from previous years, which can then be boosted by vaccines or even by infection. We need to begin to prime our immune systems to a newly emergent, related pandemic bird-flu strain, even if it is not precisely the same as that which emerges as a pandemic.
Our annual seasonal flu vaccines generally are composed of the three strains judged by WHO's expert surveillance network as most likely to spread each year. In order to begin to engender some level of immunity to bird-flu strains, we propose to add to our regular seasonal influenza immunization program a fourth strain, an H5N1 strain, after demonstrating its safety in volunteers. If this multivalent flu vaccine were efficiently deployed to all Americans, this strategy would not only prevent many of the 36,000 deaths annually in the U.S. from seasonal flu and prime our immune systems against bird strains, but also enable rapid scale-up of a specific vaccine strain when the pandemic strain emerges and create a credible market for industry.
Any new vaccine program against a pandemic raises enormous ethical and regulatory questions. If supplies were limited, who would be given vaccines? First responders? Politicians? Would enough vaccine be produced to protect people in developing countries? How would its distribution be funded? Could adjuvanted vaccines gain the same rapid FDA approval as standard seasonal vaccines to which an H5N1 vaccine could be added as simply an additional strain? Would companies have to cut back on seasonal vaccine production to create capacity for a pandemic strain? In the event of a major pandemic, would vaccination be made compulsory, as it is in almost all states for children to attend school? And how would these ethical questions be resolved in a time of crisis?
As the tragic outcomes of the 1918 flu and hurricanes Katrina and Rita compellingly tell us, our tradition of throwing money at the problem only after a disaster has occurred will not work with pandemic flu and other health emergencies. This country, as well as every city and town, needs to prepare for a pandemic in advance. We at Harvard are doing our best to contribute our knowledge and skills to this global effort--by modeling possible epidemiologic scenarios, by collaborating with colleagues in Asia, by anticipating how to respond to outbreaks of influenza, and by training public health professionals in this country and abroad to lead efforts to thwart the next pandemic.
This page is maintained by Development Communications in the Office of Resource
To contact us with suggestions, comments, and questions, please e-mail: email@example.com
Copyright, 2006, President and Fellows of Harvard College