As a low-likelihood, high-consequence event, the possibility of an influenza pandemic poses a difficult challenge to policymakers. Drawing from the ill-fated swine influenza immunization program of 1976, this article outlines 7 lessons that apply to preparations for avian influenza: (1) beware of overconfidence in models drawn from meager evidence, (2) invest in a balanced portfolio of research and contemporary preparedness, (3) clarify operational responsibilities in the federal government, (4) refrain from overstatement of objectives and misrepresentation of risk, (5) strengthen local capacity for implementation, (6) communicate strategically, and (7) lay the basis for program review.
The “swine flu affair” was an unprecedented effort in 1976 to immunize the entire US population against a possible swine influenza epidemic. The program eventually resulted in the immunization of approximately one-fifth of the US population—many more Americans than had ever been inoculated against influenza in 1 year. However, no epidemic appeared, and the reported adverse effects of vaccination led to the program's suspension. The press characterized the program as a failure and a fiasco. For public health leaders, it was a searing experience.
In 1978, at the invitation of Joseph Califano, secretary of the Department of Health, Education, and Welfare at that time, my senior colleague Richard E. Neustadt and I prepared a report on the swine influenza immunization program, which was republished with additional case material in 1983 [1, 2]. We intended this analysis to reveal lessons that could be applied to subsequent threats from influenza. We were particularly concerned with the danger that “the lessons of the crash program are learned too well—too literally—producing stalemate in the face of the next out-of-routine threat from influenza. Someday there will be one” [2, p. xxvi]. This article summarizes some of the key lessons that we gleaned from the swine flu affair and suggests ways that these lessons may be applied to preparations for avian influenza.
Features of the Swine Flu Affair
Overconfidence in theory spun from meager evidence. Major epidemics of influenza occur infrequently—perhaps 3 or 4 times/century; thus, there are relatively few opportunities for observation during a lifetime, compared with the time span available for theorizing from the data. In a review published in 1977 , Beveridge reported that the 20 major influenza pandemics between 1729 and 1968 occurred at irregular intervals of between 3 and 28 years. In contrast, on the basis of too few, relatively recent observations, leading experts in 1976 expected to encounter major influenza pandemics at regular intervals of 11 years . Similarly, the idea of antigen recycling at regular intergenerational periods of 60 years was not supported by enough data. A careful review of annual mortality due to influenza in the United States suggests caution regarding the belief that an antigenic shift is necessary for and inevitably produces a severe influenza season. In 1976, Dowdle  reported that, of the 6 peak years of excess mortality due to influenza A virus infection in the United States (1936, 1943, 1953, 1957, 1960, and 1963), only 1957 coincided with an antigenic shift in the virus. The epidemic caused by Hong Kong influenza virus in 1968 was not included in these 6 peak years and resulted in a mortality rate only slightly higher than that in 1967.
Conviction fueled by preexisting agendas. The interests and desires of many of the key players in the swine flu affair shaped their judgments. These were not matters of personal financial interest but deep-seated beliefs and goals. Jonas Salk had a long-standing interest in closing what he called the “immunity gap,” on the basis of his conviction that prevention of disease by vaccination was an achievable perfection of the human condition. Ed Kilbourne, a distinguished influenza expert, felt that active surveillance and wide-scale immunization offered a unique and valuable learning opportunity. Reuel Stallones, an adviser to the Centers for Disease Control and Prevention (CDC), saw a chance to demonstrate the value of epidemiology as a fundamental health science. David Sencer, director of the CDC, and Ted Cooper, assistant secretary for health, both believed in the central role of prevention in public health. Sencer also wished to showcase the vital place of the CDC on the national scene, and Cooper saw the program as an opportunity to promote the kind of public-private partnership that he saw as key in the advancement of the health of Americans.
Zeal by health professionals to make lay superiors “do the right thing.” The principal public health leaders felt compelled by their duty to protect the public's health. They were making a heroic response to a dramatic threat, and they were concerned about a lack of understanding among their lay superiors, who might fail to appreciate the potential catastrophe of a recurrence of the 1918–1919 pandemic. They were determined to avoid the failure to immunize sufficiently that occurred in 1957, when an outbreak began early during influenza season . They were on a mission to “do the right thing” and to make certain that their lay superiors acted accordingly.
Premature commitment. The fundamental strategic blunder of the swine flu affair was concatenating the decision to institute a universal vaccination campaign with the decision to begin manufacturing the vaccine. This premature commitment was coupled with the failure to quantify risk or to be explicit about other assumptions, such as the likelihood of a pandemic (of what size, what severity, and what duration), a sufficiently high yield from eggs used to grow the vaccine virus, protective immunity from just 1 dose of vaccine per person, high vaccine efficacy, unparalleled acceptance, favorable publicity, sustained congressional support, broad and continuing expert endorsement, widespread private involvement, adequate state operations, 3 months to complete immunization, no useful stockpiling, no liability legislation needed, and few (if any) opportunity costs. The problem expressed itself in the failure to take seriously the questions that Russell Alexander posed in early discussions at the CDC: What evidence, at what points in time, and about which things should prompt us to consider a change and to what new course?
Failure to address uncertainties. Scientists are reluctant to quantify subjective risk, and lay leaders typically refrain from eliciting quantitative estimates of risk. It is hard for anyone to separate the likelihood of an event from its severity, and this confusion distorts understanding of the meaning of “risk.” In the absence of likelihood estimates, it is difficult to be explicit about threshold conditions that might warrant a change in course. Because no lay leader had demanded any estimate of probability, Cooper could honestly tell President Ford in the summer that nothing had changed: an epidemic was still “a possibility,” even though the absence of any cases of swine influenza anywhere in the world meant that, certainly in the minds of many experts, the possibility had diminished quantitatively.
Insufficient questioning of implementation prospects. A number of the difficulties that hampered the swine influenza immunization program were not adequately understood, such as the different perspectives of public health experts, vaccine manufacturers, and insurers on the question of vaccine liability. Even with early, nearly unanimous agreement in expert opinion, opposition over time from some informed experts was to be expected. Delays in the national program arose from issues of liability, dosage, and consent. However, the most fundamental problem of implementation was the varied performance of different jurisdictions in the immunization program—a predictable problem on the basis of past immunization experiences. Cooper had a tendency to feel that he could “doctor his way through” any difficulties as they arose, as did many others.
Insensitivity to media relations and to long-term credibility. Early impressions count, and they are hard to undo, which puts a premium on preparation for the early and major news events that accompany a new national program. NBC and CBS reached different impressions at the outset of the swine influenza immunization program, because they had pursued the story in different ways—one through political contacts and the other through local contacts in Atlanta and at the CDC. In journalistic terms, each approach had merit. Ironically, the political route of inquiry led to the conclusion that the decision must be scientific, and the public health line of inquiry led to the conclusion that the decision must have a strong political component. The authorities did not adequately prepare the media for predictable events, such as coincident deaths and the possibility of previously unrecognized adverse effects (to name just 2), nor did they adequately prepare themselves to deal with contradictory views espoused by contrarian experts. This occurred in addition to the fundamental communication problem that emerges when a policy of immunization continues to be followed when the likelihood of an outbreak declines. By not laying the basis for program review and not establishing review checkpoints, public officials placed their institutions' long-term credibility at risk. They concentrated on the worst case in the short run—namely, a severe pandemic. If they had thought as hard about the most likely case in the long run (adverse effects, lawsuits, and no pandemic), the risk of diminished credibility would have loomed large.
Lessons From the Swine flu Affair and Avian Influenza Preparedness
A number of parallels between the swine flu affair in 1976 and avian influenza in 2008 are obvious; others are subtle. In the attempt to identify lessons, dissimilarities are as important as similarities . The specter of the great influenza pandemic of 1918–1919 cast a harsh shadow in 1976 and 2008. The threat of a potential pandemic among humans spurred preparations in both years. Although scientific understanding of influenza has advanced strikingly over the past 3 decades, uncertainties regarding likelihood and severity have been a shared reality in both years. National immunization programs depend profoundly on the success of preparation and execution at the local and state levels. Public understanding and support remain critical to a successful program. Among the more salient distinctions between then and now are the global scope and distribution across species of avian influenza and the consequent need for global cooperation; the absence of an available vaccine to protect human populations from avian influenza; the current existence of antiviral medication as a potential tool for the prevention of or response to a pandemic; the complexity of governmental organization and decision making in post—September 11 America; and the novel and diverse forms of public media and communication, including e-mail, blogs, Web sites, cable channels, and cell phones with instant messaging.
In the case of swine influenza, the preoccupying policy question was whether to embark on a mass immunization campaign, which then led to the associated matters of vaccine production, field trials, program implementation, media relations, and surveillance. In the case of avian influenza, the policy challenges are more diverse and global: worldwide surveillance and laboratory capacity, including cooperation and transparency across national boundaries; the management of animal outbreaks, the migration of and trade in animals for food and pets, and human- animal interactions; the value and deployment of an array of nonpharmacological infection-control measures, ranging from face masks to travel restrictions, quarantine and isolation, and school closures; the availability and role of antiviral medication in the prevention of and response to disease outbreaks, including international collaboration and aid; the preparedness of hospitals and health care providers for a possible onslaught of patients, including the availability of respirators and infection-control facilities; the potential disruption of essential services, economic dislocation, and the maintenance of civil order; and research strategy and timing for new vaccines and field trials, for more-efficient vaccine-production methods and increased production capacity, and for rapid, accurate, field-based diagnostic tests. The US Congress has enacted legislation (Public Readiness and Emergency Preparedness Act, Public Law 109–148, 30 December 2005) that provides liability protection to vaccine manufacturers during a declared public health emergency; thus, one stumbling block from the past has been removed.
Bearing in mind the similarities and differences between swine influenza and avian influenza, I suggest 7 lessons for avian influenza preparedness.
Beware of overconfidence in models drawn from meager evidence. Today's model builders, like the influenza theorists a generation ago, can lead the unwary into overconfident expectations. Properly interpreted, models play a valuable role. Models can help identify the requirements needed for interventions to succeed, clarify thinking about alternatives and their consequences, enable the exploration of particular assumptions and structural elements without losing sight of the whole, and detect variables in which relatively small changes can strongly influence results. The model builders can quantify the degree of uncertainty that should be considered with their predictions, which can help deter overinterpretation. Models can produce more-robust conclusions if they include a variety of structural assumptions and ranges of estimates and if they are supplemented with information from a variety of sources, such as historical case studies, laboratory findings, and field investigations. The most useful models will incorporate sensitivity analyses, measures of uncertainty, and changes in assumptions; provide indicators of costs and burdens; and be designed to adapt in real time to new evidence from actual experience.
Invest in a balanced portfolio of research and contemporary preparedness. Dollars spent on key research areas, such as the development of a safe and effective vaccine and of rapid diagnostics and understanding the molecular foundation of virus transmission, infectivity, and virulence, are necessary to improve the capacity of the United States and the world to respond to a pandemic threat. These investments are best understood as a trade-off between present preparedness and stronger future capacity to prevent and cope with a pandemic. Today, the technology for production of influenza vaccine is essentially the same as that in the 1970s; this must not be permitted to be true for another generation. Regardless of whether the current avian influenza A(H5N1) virus causes the next pandemic among humans, the next worldwide influenza pandemic will occur someday.
Clarify operational responsibilities in the federal government. The post—September 11 dispersal and redefinition of authority over the management of a potential influenza pandemic is the most striking difference in the national policymaking landscape, compared with that in the mid-1970s. It is inconceivable today that the CDC should have sole responsibility for assessment of the situation and formulation of the policy response. In 1976, the assistant secretary for health had line authority over the several health-related agencies in the Department of Health, Education, and Welfare (i.e., the CDC, the National Institutes of Health, and the Food and Drug Administration). Today, these agency heads report directly to the secretary of Health and Human Services, and the assistant secretary for health serves in a senior staff capacity, along with a separate senior department official who is responsible for bioterrorism preparedness. More fundamentally, today's Department of Homeland Security changes the balance of authority among all departments in the federal government, including the Department of Health and Human Services. Because the Department of Homeland Security has jurisdiction over natural disasters and national emergencies and the Department of Health and Human Services has jurisdiction over health matters, there is room for uncertainty and bureaucratic jockeying over responsibility for the response to a health emergency, such as an influenza pandemic. Is it to be regarded mainly as a national emergency, with the Department of Homeland Security in the lead, or as a health problem, with the Department of Health and Human Services in the lead? Recent legislation (Pandemic and All-Hazards Preparedness Act, Public Law 109–417, 19 December 2006) directs the secretary of Health and Human Services to lead all federal public health and medical responses to public health emergencies. Although this is an important clarification, there will be a need for ongoing interdepartmental dialogue and coordination that involves the Departments of Health and Human Services, Homeland Security, State, Defense, Education, and Transportation, among others, and that includes the White House Homeland Security Council and appropriate state-level authorities and agencies.
Refrain from overstatement of objectives and misrepresentation of risk. Communication about risk is difficult, because the public, like many experts, has a hard time separating likelihood from severity. Estimation of risk is difficult when the observable instances are widely separated in time, as with pandemic influenza. Explaining the concept of uncertainty as an estimate of likelihood is challenging. As the scientific capacity for global surveillance improves, distinguishing between when a particular occurrence is happening for the first time and when it merely is being detected for the first time will be difficult. Eventually, research at a molecular level may lead to a high level of confidence in our knowledge of the determinants of transmissibility and disease severity, which could improve future assessments. In the meantime, setting the right balance regarding what is known about risk will be a challenge, and it can be expected that, over time, extreme positions of expectation (“must occur” or “surely will not occur”) will be espoused. When goals for a program are announced publicly, setting achievable results based on honest assessments of capacity will pay dividends in credibility over the long term. However, this is not an excuse for settling for an inadequate capability to deliver results.
Strengthen local capacity for implementation. The greatest vulnerability to success in 1976 remains present in 2008—namely, the ability of states and communities to implement a program. We will not have a national plan for avian influenza until there are 50 state plans and thousands of local plans for every jurisdiction in the United States. In addition, plans are not enough. Exhortation of states and cities is not enough. Once we have a vaccine, what will it take in every community to be able to immunize 250 million American residents—that is, five-sixths of the population—in 6–7 weeks? What will it take to effectively distribute antiviral medications, make face masks available, or function with closed schools and businesses? The speed and scale of the action required may be unprecedented. The seasonal hodgepodge of routine influenza vaccination that includes clinics, doctors' offices, hospitals, work sites, and pharmacies, each available on occasional and uncoordinated days, simply is not up to the task. In what town or city has every household received a card telling them where and when to go to get what they need in the case of an emergency? If past performance is any guide, variation in local preparedness and delivery capacity remains the key shortcoming in the national readiness to cope with a major pandemic. The responsibility to upgrade delivery capacity is shared at local, state, and national levels. Federal-level agencies and officials could help by specifying evidence-based guidelines and standards in detail, identifying and sharing best practices, providing technical assistance to achieve acceptable levels of performance, and ensuring the availability of physical and financial resources to each state.
Communicate strategically. Media standards and values differ from those of the scientific and health communities. Public Editor Byron Calame of the New York Times recently described 7 goals of reporters . To paraphrase, these goals are be first, write stories with impact, win prizes, impress sources, figure out what is really happening, tell stories in a compelling way, and get on the front page. These aims do not coincide with the health expert's goal of educating the public and gaining public confidence, understanding, and cooperation. For example, a newspaper or magazine editor can claim to have portrayed an issue “accurately” when the publication has quoted correctly an “expert” who espouses a scientifically absurd position. The media thrive on controversy, and public health officials can expect to see those with contrary views quoted in the media, even (or especially) if they are in a minority. Strategic communication means mapping the message, the audience, the messenger, and the medium onto one another, to attain the intended effect in the intended audience. The advent of new electronic media (Web sites, both official and unofficial; e-mail; and blogs), cell phones with instant messaging, and the fractionation of the broadcast space by cable and satellite create an immensely more-varied and challenging communication environment. These developments have created an added premium for clear leadership, knowledgeable spokespersons, and sophisticated outreach through new media as well as traditional media. In a pandemic situation, public understanding and cooperation will count for a great deal.
Lay the basis for program review. Decision making in advance and during an influenza pandemic will require a sequence of choices that, ideally, will accommodate new and unexpected developments. Many past influenza pandemics introduced unexpected features, such as high mortality among young adults in 1918–1919 and early outbreaks in 1957 (to name just 2). Laying the basis for program review means being explicit about assumptions and scrutinizing their foundations, which will require sensitivity to well-meaning personal agendas that shape perspective, as well as to the institutional and bureaucratic impulses to control and manage. It also will require candor about the limitations of expert knowledge and a recognition that often what you do not know will not hurt you nearly as much as what you confidently believe to be true but just is not. Laying the basis for program review means being quantitative about likelihoods and expectations, and it means establishing checkpoints and taking Alexander's questions seriously: What evidence, at what points in time, and about which things should prompt us to consider a change and to what new course? A worthy program will have leaders who are prepared to incorporate new information in real time and to make decisions and act accordingly.
Policymaking for avian influenza preparedness is problematic in part because an influenza pandemic is a low-likelihood, highconsequence event. In such cases, steps toward preparedness are subject to criticism as both unnecessary (in the likely case of no event) and inadequate (if a catastrophic event occurs). This politically precarious double bind reinforces the value of learning the strategic lessons from past errors of over- and underreaction and applying them to the realities of today.
I dedicate this article to the memory of Richard E. Neustadt. The Harvard University Asian Flus and Avian Influenza Workshop was hosted by the Harvard University Department of Anthropology, Harvard School of Public Health, and Harvard Asia Center and was supported by the National Science Foundation, Harvard Asia Center, and the Michael Crichton Fund.
Supplement sponsorship. This article was published as part of a supplement entitled “Avian and Pandemic Influenza: A Biosocial Approach,” sponsored by the National Science Foundation, Harvard Asia Center, and the Michael Crichton Fund.