A Potential Influenza Pandemic: Possible Macroeconomic Effects and Policy Issues

CONGRESSIONAL BUDGET OFFICE Douglas Holtz-Eakin, Director

U.S. Congress

Washington, DC 20515

www.cbo.gov

December 8, 2005; revised July 27, 2006

Honorable William H. Frist, M.D.

Senate Majority Leader

United States Senate

S-230 Capitol

Washington, DC 20515

Dear Senator Frist:

In response to your request, the Congressional Budget Office (CBO) has prepared an assessment of

the possible macroeconomic effects of an avian flu pandemic. The assessment concludes that a

pandemic involving a highly virulent flu strain (such as the one that caused the pandemic in 1918)

could produce a short-run impact on the worldwide economy similar in depth and duration to that of

an average postwar recession in the United States. Most of the pandemics of the past involved much

milder strains; an outbreak of that kind would have a much smaller economic impact, which might be

indistinguishable in the macroeconomic data.

In its assessment, CBO also describes the current state of preparedness for addressing a possible

pandemic and options for increasing the nation’s preparedness. (However, in accordance with CBO’s

mandate to provide objective, impartial analysis, this document contains no recommendations.) The

billions of dollars spent in recent years preparing for health crises that could arise from possible

terrorist attacks would provide some limited benefits in the event of a flu pandemic. Even so, if a

pandemic were to occur in the near term, the options for the United States would be limited to

attempts to control the spread of the virus and judicious use of limited medical facilities, personnel,

and supplies. In the longer term, more tools are potentially available, including an increased treatment

capacity, greater use of vaccines and antiviral drug stockpiles, and possible advances in medical

technology.

The details of CBO’s assessment are in the attachment to this letter. We would be pleased to address

any further questions you might have.

Sincerely,

Douglas Holtz-Eakin

Attachment

cc: Honorable Harry Reid

Minority Leader

United States Senate

Honorable Nancy Pelosi

Minority Leader

U.S. House of Representatives

Honorable J. Dennis Hastert

Speaker

U.S. House of Representatives

The Congress of the United States O Congressional Budget Office

A Potential Influenza Pandemic:

Possible Macroeconomic Effects and

Policy Issues

December 8, 2005; revised July 27, 2006

This revision incorporates corrected estimates of the supply-

side impact of a potential pandemic: about 2-1/4 percent in the

severe scenario and about 1/2 percent in the mild scenario. The

estimated overall impact on gross domestic product is about

4-1/4 percent in the severe scenario and about 1 percent in the

mild scenario.

Note

This assessment was written by Robert Arnold, Jeanne De Sa, Tim Gronniger,

Allison Percy, and Julie Somers under the supervision of Robert Dennis, Joseph Kile,

David Moore, and Robert Sunshine. Outside the Congressional Budget Office

(CBO), the following individuals read an early draft and provided technical advice

and helpful comments. However, the assistance of external reviewers implies no

responsibility for the final product, which rests solely with CBO.

Marc Lipsitch, D.Phil., Department of Epidemiology, Harvard School of

Public Health

Pamela McInnes, D.D.S., M.Sc. (Dent.), Deputy Director, Division of

Microbiology and Infectious Diseases, National Institute of Allergy and

Infectious Diseases

Christina Mills, Sc.D. Candidate, Department of Epidemiology, Harvard

School of Public Health, and M.D. Candidate, Harvard Medical School

William D. Nordhaus, Ph.D., Professor of Economics, Yale University

Peter Palese, Chair, Department of Microbiology, Mt. Sinai School of

Medicine

Harvey Rubin, M.D., Ph.D., Director, Institute of Strategic Threat

Analysis and Response, University of Pennsylvania

iii

Contents

Introduction and Summary 1

Background 2

Possible Pandemic Scenarios 6

Economic Effects of a Pandemic 9

Short-Term Effects 10

Long-Term Effects 15

Current Policies and Options for the Future 16

Context 17

The Administration’s Proposal and Congressional Alternatives 18

Options for the Longer Term 32

Technical Appendix 41

Tables

A-1. Assumptions Underlying Estimates of the Supply-Side

Impact of an Avian Flu Pandemic 42

A-2. Assumed Declines in Demand, by Industry, in the

Event of an Avian Flu Pandemic 44

Boxes

1. Problems in the Market for Influenza Vaccine 24

2. “Surge” Medical Capacity of the Departments of Defense,

Veterans Affairs, and Homeland Security 30

3. Obstacles to Increasing the Production of Currently

Available Antiviral Drugs 38

Introduction and Summary

There is widespread concern among policymakers and public health experts about the

possibility of a worldwide epidemic of avian influenza. Such pandemics are not new:

there were three in the 20th century, of which one, the 1918–1919 Spanish flu out-

break, is estimated to have killed over 500,000 people in the U.S. and up to 50 mil-

lion worldwide. Public health concerns arise because of the challenge of creating the

public health infrastructure in the United States and other countries that would be ad-

equate to meet the challenges of a severe pandemic.

Although a pandemic could be caused by any of several influenza strains, scientists are

particularly worried about H5N1, a strain that has caused repeated epidemics with

high mortality among poultry in Asia, has spread from Southeast Asia to flocks in

Central Asia and Europe, and has made the jump from birds to humans, causing the

deaths of over 60 people. Moreover, viruses of the H5 subtype are not known to have

ever circulated among the human population, which means that there would be little

immunity to it. To date, close contact with infected poultry is thought to be required

for human infection, but the danger exists that the virus will evolve in a way that al-

lows for efficient human-to-human transmission. If the virus does acquire that capa-

bility, a worldwide epidemic, or pandemic, could occur. Depending on the virulence

of the particular strain of flu, such an outbreak could have substantial consequences

for people and economic activity around the world.

Infectious diseases are, however, unpredictable. It is impossible to say for sure whether

another pandemic will arise, whether it will involve H5N1, and, if it does, when it

will happen or whether it will be mild or severe. The H5N1 virus could mutate in a

way that caused a severe pandemic next year or a mild epidemic in a decade or two.

Or it could evolve in a way that rendered it harmless, and a pandemic could arise from

an entirely different virus subtype.

This paper focuses on the potential for, and possible economic effects of, a pandemic

of avian flu—although many of the policy issues that avian flu raises apply also to

pandemics of other types of influenza. The paper provides background on the effects

of a potential avian flu pandemic, gives very rough estimates of the economic effects

of two possible scenarios, and discusses policy options related to the preparedness of

the United States for such an outbreak.

Based on an analysis of past pandemics, CBO has devised two scenarios to outline the

possible economic effects of a potential avian influenza pandemic. There is a substan-

tial amount of uncertainty associated with these scenarios because there is scant em-

pirical evidence available to inform many of the assumptions that are needed for the

calculations underlying the economic effects. The first, and more severe, scenario is

roughly similar to the 1918-1919 Spanish flu outbreak. In the severe-pandemic sce-

nario, roughly 90 million people become sick and 2 million people die in the United

States, and in CBO’s estimation, real GDP would be about 4-1/4 percent lower over

the subsequent year than it would have been had the pandemic not taken place. That

estimate of the effect on GDP is comparable to the effect of a typical business-cycle

recession in the United States during the period since World War II. In the mild-pan-

demic scenario, which resembles the 1957 and 1968 pandemics, about 75 million

people are infected in the United States and about 100,000 of them die. In that sce-

nario, the pandemic reduces real GDP by a modest amount, about 1 percent relative

to what would have happened without a pandemic, but probably would not cause a

recession and might not be distinguishable from the normal variation in economic

activity.

The Administration and the Congress have proposed to improve preparedness for a

potential pandemic. But the uncertainties associated with whether a pandemic flu will

occur in the next several months or in the next five years—and if an outbreak does oc-

cur, what measures will be effective in reducing its human and economic costs—make

finding the best approach a difficult task. Improving the capacity of the health system

to care for many people in all parts of the country who are sick at the same time

stands out as a priority—but one that the federal government may be less suited to

address because of the local delivery of health care services. A promising federal object

of attention is support for the efforts of governments of other countries and interna-

tional organizations to contain avian influenza strains and control their evolution to

diseases that are easily transmitted from person to person. Federal policies may also be

effective in building stockpiles of vaccines and improving antiviral drugs as well as

putting in place new technologies that allow effective vaccines to be produced more

rapidly and in large quantities. The challenge for vaccine, drug, and technology policy

will be to do enough to provide private producers with incentives to meet national

needs without introducing inefficiencies that have sometimes occurred when govern-

ment is very actively involved in specific decisions about technologies and invest-

ments.

Background

Avian influenza (or “bird flu”) is a contagious animal disease that infects birds and

some mammals.

Scientists believe that all bird species are susceptible to infection but

that some are more resistant than others. Wild waterfowl, especially ducks, are a so-

called natural reservoir of influenza viruses, including the bird flu. The birds carry the

virus without displaying any symptoms of the disease and can spread the virus over

great distances while remaining healthy.

Poultry are quite susceptible to avian influenza, which can cause a wide range of

symptoms, from mild (reduced egg production) to severe (rapid death). The severe

1. Material for the background section of this paper was drawn from several sources but primarily

World Health Organization, Avian Influenza: Assessing the Pandemic Threat (Geneva: World

Health Organization, January 2005); Howell Pugh, “Pandemic, The Cost of Avian Influenza,”

Contingencies (September/October 2005), pp. 22-27; and Laurie Garrett, “The Next Pandemic,”

Foreign Affairs (July/August 2005), pp. 3-23.

form of the disease, which is known as “highly pathogenic avian influenza,” is ex-

tremely contagious and has been the source of numerous epidemics among domesti-

cated birds. It is also characterized by very high and rapid mortality, with rates ap-

proaching 100 percent and death sometimes occurring on the first day that symptoms

appear.

Although frequently deadly for poultry, avian influenza viruses in the past have rarely

caused severe disease in humans. However, in 1997, a highly pathogenic strain of bird

flu known as H5N1 jumped from birds to humans during an outbreak among poul-

try in Hong Kong.

The 1997 event was notable for two reasons. First, molecular

studies indicated that the genetic makeup of the human and avian viruses were virtu-

ally identical, indicating direct transmission from birds to humans.

Second, the

H5N1 virus caused severe illness with extreme mortality among humans: of the 18

persons known to be infected, six died. The outbreak ended after authorities slaugh-

tered Hong Kong’s entire stock of poultry (about 1.5 million birds).

The Hong Kong episode put world health officials on alert because the H5N1 strain

had fulfilled two of the three prerequisites for a pandemic. First, the strain was a new

virus subtype to which the population would have little or no immunity, and second,

the virus had the ability to replicate in humans and cause serious illness. However, the

virus still has not developed the ability to be transmitted efficiently from human to

human.

The genetic makeup of influenza viruses can change in two different ways. The first

type of change, known as antigenic drift, occurs when small errors creep into a virus’s

genetic sequence when the virus copies itself. As a result, influenza viruses undergo

frequent small changes in structure, or mutations, which allow them to continually

find susceptible populations to infect. The second type of change involves a swapping

or mixing of genes between two types of virus—such as avian and human forms of in-

fluenza—when they are present in the same animal or individual. That change is

termed reassortment, or antigenic shift. If either type of genetic change produced an

influenza strain with a particular mix of genes—one that caused severe disease and al-

lowed for efficient transmission among humans—it could ignite a pandemic.

2. Strains of influenza are grouped into subtypes that are identified by the hemagglutin (HA) and

neuraminidase (NA) protein spikes on their surfaces. A population that has been exposed to a

strain acquires “herd immunity” when enough individuals have gained immunity to interfere with

the transmission of the virus. Of the 15 HA subtypes, H1, H2, and H3 are known to have circu-

lated among humans during the past century. Hence, a virus from the H5 subtype, such as H5N1,

would probably be unfamiliar to the immune system of everyone in the world, and there would be

no herd immunity.

3. More commonly, avian influenza’s route from birds to humans involves an infection in a third spe-

cies, usually thought to be pigs (because they are susceptible to both human and avian influenza),

during which genetic material is exchanged.

Since the 1997 episode in Hong Kong, there have been several outbreaks of the

H5N1 influenza around the world, leading to tens of millions of infections among

poultry and dozens of cases among humans. The first human infection occurred in

Hong Kong in February 2003, when a nine-year-old boy and his father became sick

after a trip to southern China. The man, who was 33, died, but his son recovered.

Then, in 2004, the H5N1 virus spread among poultry populations in Southeast Asia,

with outbreaks of influenza reported in two separate waves. The first wave, in January

and February, affected Vietnam, Japan, Korea, Thailand, Laos, Cambodia, Indonesia,

and China. The second wave, which began in July and continued into 2005, included

outbreaks in the same countries and in Malaysia as well. More recently, the virus has

shown up in Russia, Kazakhstan, Turkey, and Romania. The scope of the outbreaks is

historically unprecedented: a highly pathogenic strain of avian influenza has not been

known before to spread so widely and so rapidly.

The number of human cases of the H5N1 virus has also grown. Between January

2004 and August 2005, there were 112 human cases of H5N1 avian flu (in Vietnam,

Thailand, Cambodia, and Indonesia) that resulted in 57 deaths. The vast majority of

those cases (and deaths) involved children and young adults. However, accurately

computing the case fatality rate (the percentage of infected persons who eventually die

from the disease or from its complications) is impossible because authorities do not

know how many people had milder cases but did not seek medical care or how many

received care that was not reported. Nearly all of the human cases resulted from close

contact with infected birds. There is evidence, though, of at least one case of probable

human-to-human transmission, and some experts suspect that a few other cases of hu-

man-to-human spread of the H5N1 virus have occurred.

Events since the beginning of 2004 have heightened concerns among public health of-

ficials. Not only has the H5N1 virus spread widely—expanding beyond Southeast

Asia and China into Central Asia and Europe—but laboratory results indicate that

the virus has evolved in ways that may make a pandemic more likely:

B It has found a permanent ecological niche, becoming entrenched among domestic

ducks in rural areas of Asia.

B It has become more robust than the 1997 strain and is able to survive longer under

a broad range of environmental conditions.

B It has become increasingly pathogenic in poultry and has increased the range of

species it can infect, now including domestic cats (in laboratory experiments) and

captive tigers (after being fed infected chicken carcasses in a zoo in Thailand).

4. See World Health Organization, “Avian Influenza Frequently Asked Questions,” available at

http://www.who.int/csr/disease/avian_influenza/avian_faqs/en/; and Centers for Disease Control

and Prevention, “Information About Avian Influenza (Bird Flu) and Avian Influenza A (H5N1)

Virus,” CDC Fact Sheet (May 24, 2005).

B It has become resistant to one of the two classes of antiflu drugs.

Experts do not know if an avian influenza pandemic is likely to occur, largely because

they cannot predict when, or even if, the H5N1 virus might acquire the ability to pass

readily from human to human. But the wider presence of the avian strain raises the

probability of a pandemic because it increases the likelihood that an individual will

become infected with the human strain and the avian strain at the same time, thus

opening up the possibility of a genetic reassortment that could improve transmissibil-

ity of the disease. Wider prevalence of the virus also increases the likelihood that a se-

ries of mutations will produce a pandemic strain, even without the virus’s undergoing

reassortment.

Nevertheless, despite the widespread concern, some scientists caution that an H5N1

pandemic is not a forgone conclusion.

Although they agree that a pandemic is possi-

ble, those researchers argue that the H5 subtype of the influenza virus has not shown

the ability to pass efficiently among mammals. Indeed, one mystery associated with

the H5N1 virus is why it has not reassorted, despite ample opportunity to do so. Hu-

man and avian strains have circulated concurrently since 1997 (and perhaps earlier),

and thousands of workers were exposed to H5N1 during poultry-culling operations

in Asia. Luck may have played a role, but another explanation is that H5N1 may not

possess the capacity for efficient human transmission.

Although no scientific basis exists to calculate the probability of an influenza pan-

demic during the next 10 years, history provides a gauge to judge those chances at

least roughly. Since 1700, there have been between 10 and 13 influenza pandemics (or

probable pandemics) in the world, including the three that have occurred since the

beginning of the 20th century.

Those pandemics have taken place at irregular inter-

vals, with as little as two years separating some outbreaks and as many as 55 years sep-

arating others. Hence, one could assume that there is a roughly 3 percent to 4 percent

probability of a pandemic occurring in any given year. Gauging the severity of a po-

tential pandemic is more difficult. Of the three pandemics that have occurred since

the beginning of the 20th century, two were mild and one was severe, suggesting a

fairly high probability of a severe event. However, the 1918 pandemic appears to be

unique in its severity when compared with the other post-1700 pandemics. For exam-

ple, although the estimates are very rough, case fatality rates during the pandemics in

the 18th and 19th centuries more closely resembled those of the mild pandemics of

5. See Dennis Normile, “Pandemic Skeptics Warn Against Crying Wolf,” Science, vol. 310 (Novem-

ber 18, 2005), pp. 1112-1113.

6. That argument relates solely to H5-subtype viruses. Researchers who believe that an H5N1 pan-

demic is unlikely allow for the possibility of a pandemic caused by an H1, H2, or H3 subtype.

7. According to Michael Osterholm (“Preparing for the Next Pandemic,” Foreign Affairs, July/August

2005), there have been 10 influenza pandemics during the past 300 years, while K. David Patter-

son (Pandemic Influenza 1700-1900: A Study in Historical Epidemiology, Totowa, N.J.: Rowman &

Littlefield, 1986) asserts that there have been 13 pandemics since 1700.

1957 and 1968 than those of the 1918 pandemic. Moreover, mortality was generally

concentrated among the elderly, and the degree of social disruption was much smaller

than during the 1918 episode.

Consequently, based only on historical frequencies,

the chances of a severe pandemic’s occurring appear to be relatively small, perhaps on

the order of 0.3 percent. However, given the evidence of an existing epidemic of

H5N1 in fowl, and the possibility that it might mutate to circulate efficiently in hu-

mans, the probability of a severe pandemic may exceed the historical frequency.

Possible Pandemic Scenarios

The uncertainty surrounding the occurrence of an influenza pandemic makes it very

difficult to forecast the economic effects of such an event. As noted earlier, scientists

do not know whether the H5N1 virus will acquire the properties necessary for effi-

cient human-to-human transmission. Moreover, even if the virus acquired those prop-

erties, there is no way to predict how virulent the resulting strain would be. Conse-

quently, a substantial amount of uncertainty is associated with any forecast of the

effects of an avian flu pandemic. The Congressional Budget Office (CBO) has devised

two scenarios based on past pandemics to outline possible outcomes.

The 20th century has seen three influenza pandemics: the 1918-1919 Spanish flu out-

break, the Asian flu pandemic of 1957-1958, and the 1968-1969 outbreak of Hong

Kong flu.

Of the three, the 1918-1919 pandemic was the most severe. An estimated

25 percent to 30 percent of the world’s population became ill, and the death rate is es-

timated to have been as high as 11 percent of those stricken. Estimates of deaths as a

result of the Spanish flu range from 500,000 to 675,000 in the United States and

from 40 million to 50 million worldwide.

(However, estimates of the number of

persons affected are extremely rough because a clinical test for influenza had not been

developed at that time.) The 1918 strain of influenza was not unusual with regard to

its transmissibility, but it was unusual in its very high rate of mortality and in the large

percentage of deaths among victims between the ages of 15 and 35. Most deaths from

influenza are caused by pneumonia resulting from a secondary bacterial infection. Al-

though many of the deaths in 1918 and 1919 followed that pattern, the Spanish

flu pandemic was unusual in that a large proportion of the deaths were caused

8. See Patterson, Pandemic Influenza 1700-1900, p. 91, who stresses the unique nature of the 1918

outbreak. No other pandemic spread so rapidly, killed as many victims, or caused anything like the

mortality seen among young and middle aged people.

9. Why the virus that caused the 1918 pandemic was termed the “Spanish” flu is unclear, given that

there is no evidence that it originated in Spain or that it hit that country particularly hard. Possi-

bly, the term arose because of heavy coverage by the Spanish media.

10. Deaths attributed to past pandemics reflect mortality above that expected in a normal flu season.

According to the Centers for Disease Control and Prevention, 5 percent to 20 percent of the pop-

ulation catches the flu each year in the United States, and roughly 36,000 people die from the

disease.

directly by the virus and the body’s immune response to it, which could kill young,

otherwise healthy, people in less than 48 hours.

The pandemics of 1957 and 1968 were much milder than the Spanish flu outbreak.

Part of the reason was that the influenza strains in those later pandemics were less vir-

ulent, but there were also other contributing factors. Medical technology had pro-

gressed, and, in the case of the 1968 episode, some immunity had been conferred by

exposure to the 1957 strain. The spread of the viruses in the two events was character-

istically rapid, but the lethality of the two strains was much less than that of the 1918

virus. In addition, global surveillance had improved, which allowed public health au-

thorities to quickly isolate the viruses, and manufacturers were able to provide vac-

cines for the two strains before the pandemics had eased. Also lessening the later pan-

demics’ severity was that doctors were able to use antibiotics and other post-1918

technology to deal with bacterial infections and there were fewer cases of viral pneu-

monia than in 1918.

Perhaps as a result, the pattern of mortality in the 1957 and 1968 pandemics was

more typical of normal seasonal outbreaks (that is, it was concentrated among the

very young and very old). The two outbreaks killed far fewer people who previously

had been healthy. The 1957 outbreak is estimated to have caused 2 million deaths

worldwide and roughly 70,000 in the United States. The 1968 pandemic is thought

to have caused 34,000 deaths in the United States. The viruses that caused the pan-

demics of 1957 and 1968 are known to have been caused by genetic reassortment, or,

an exchange of genes between avian and human influenza viruses.

Although the dimensions of a future flu pandemic are unknowable, past outbreaks

suggest the following pattern of events:

B The virus would spread widely in a very short time. On the basis of experience with

severe acute respiratory syndrome (SARS) in 2003, a pandemic influenza virus

would be expected to cross national borders very rapidly.

B A rapid surge in the number of cases in each affected area would occur very quickly,

within weeks. The number of cases would vary with the severity of the outbreak,

but there would be a sharp increase in demand for medical services.

11. The H5N1 strain of avian influenza has shown a similar ability to cause a fatal form of viral pneu-

monia in human cases.

12. In contrast, the virus that caused the 1918 pandemic is thought to have evolved from a purely

avian strain through a series of mutations (“antigenic drift”) that produced a genetic structure that

allowed for efficient transmission among humans.

B The pandemic would probably spread across geographic areas and vulnerable pop-

ulations in waves. In any given geographic region, each wave could last for three to

five months, and a second wave could appear anywhere from one to three months

after the first disappears.

For this analysis, CBO developed two pandemic scenarios, one mild and one severe,

which it based on past experience, applying the infection and case fatality rates from

the earlier pandemics to the current U.S. population of just under 300 million. Esti-

mates of the percentage of the population that became ill with influenza during past

pandemics (the so-called gross attack rates) are extremely rough. But experts generally

agree that the three past outbreaks during the 20th century did not differ markedly

with respect to their attack rates and that those rates ranged from about 25 percent to

30 percent. Consequently, CBO applied a 25 percent attack rate in the mild scenario

and a 30 percent rate in the severe scenario.

The reason that the 1918 pandemic was so much more severe than the two later out-

breaks was that it killed a far greater percentage of those that it infected than did the

1957 or 1968 strains. Like the attack rate, the case fatality rate for a pandemic is ex-

tremely difficult to estimate, largely because the total number of infected people, in-

cluding those who do not seek treatment, is unknown. Nonetheless, using available

evidence, experts estimate that the case fatality rate during the 1918 outbreak was

about 2.5 percent in the United States, whereas during the 1957 and 1968 episodes,

the case fatality rate ranged from just under 0.1 percent to about 0.2 percent.

CBO

thus assumed a rate of 2.5 percent for the severe scenario and just over 0.1 percent for

the mild scenario.

Using those assumptions, the two pandemic scenarios that this paper considers are:

B A severe pandemic, similar to the 1918-1919 episode, that could infect 90 million

people in the United States and cause the deaths of more than 2 million of them;

and

B A mild pandemic that resembled the 1957 and 1968 outbreaks, which might be

expected to infect 75 million people and cause roughly 100,000 deaths in the

United States.

13. During the 1918-1919 pandemic, the second wave (which began at the end of August 1918) had

a higher case fatality rate than did the first (which had begun the previous spring). During the

1957 pandemic, the first wave primarily affected school-age children, whereas the second had a

greater impact among the elderly.

14. For more details on attack and case fatality rates, see Martin Meltzer, Nancy Cox, and Keiji

Fukuda, “The Economic Impact of Pandemic Influenza in the United States: Priorities for Inter-

vention,” Emerging Infectious Diseases, vol. 5, no. 5 (September-October 1999).

Both scenarios presume that effective vaccines are not available in time to significantly

alter the pandemic’s course. Quarantine and travel restrictions are among the other

possible policy responses, but those approaches have been shown to have little effect

on overall mortality and only a limited ability to forestall the onset of local epidemics.

Although it is impossible to predict with confidence what the next pandemic will look

like, several factors suggest that the worst-case scenario will be less severe than the

1918 pandemic outlined earlier. Medical technology has advanced significantly, pro-

viding health care providers with more information and better treatment options, es-

pecially for complications associated with bacterial infections. Antiviral drugs are also

available and, if provided quickly, offer the prospect of decreasing the severity of infec-

tion.

In addition, international mechanisms have been put in place that allow for

better surveillance and a more rapid response to a new disease. Once the virus had

been identified, vaccines would be developed to protect vulnerable populations, an

option that was not available in 1918. However, the length of time required to pro-

duce sufficient quantities of a vaccine would limit its ability to lessen the effects of the

pandemic.

Balanced against those factors are some that might suggest a worse outbreak than the

one that occurred in 1918. The world is now more densely populated, and a larger

proportion of the population is elderly or has compromised immune systems (as a re-

sult of HIV). Moreover, there are interconnections among countries and continents—

faster air travel and just-in-time inventory systems, for example—that suggest faster

spread of the disease and greater disruption if a pandemic was to occur.

Economic Effects of a Pandemic

Just as it is difficult to forecast the severity of a pandemic, it is hard to predict its eco-

nomic effects, even if the outbreak’s scope and severity are known. Only a few esti-

mates based on past flu epidemics exist, so this paper will rely in part on evidence

from the SARS epidemic of 2003. Based on past influenza pandemics and the SARS

outbreak, the most important effects would be a sharp decline in demand as people

avoided shopping malls, restaurants, and other public spaces, and a shrinking of labor

supply as workers became ill or stayed home out of fear or to take care of others who

were sick.

An avian flu pandemic could be thought of as a “shock” to the economy, with both

demand- and supply-side effects in the short run. In addition, the pandemic would

have longer-term supply-side effects. The short-term effects of the pandemic would

depend on its scope. Under the conditions in the severe scenario described above, the

15. For evidence on that point, see Neil Ferguson and others, “Strategies for Containing an Emerging

Influenza Pandemic in Southeast Asia,” Nature (September 8, 2005), pp. 209-214; and Ira

Longini and others, “Containing Pandemic Influenza at the Source,” Science (August 12, 2005),

pp. 1083-1087.

human toll would be devastating, and the economic effects would be greater than in

recent recessions and roughly the same size as the average postwar recession. In the

long term, however, the economy’s response to natural disasters demonstrates that

people can adapt to extreme hardship and businesses can find ways to work around

obstructions. As a result, economic activity would recover, and the economy would

eventually return to its previous trend growth rate.

What follows will be a discussion of the economic effects of an influenza pandemic,

with emphasis on the reaction to a severe outbreak. Many of the short-run disruptions

to the economy that would come to pass under the conditions of CBO’s severe sce-

nario might also occur in the event of a relatively mild outbreak. That is, the public’s

response to an avian flu pandemic, as to the SARS outbreak, might be disproportion-

ate to the event’s clinical severity (or lethality). However, it might not be clear during

the outbreak that the public’s response was disproportionate because there would be a

considerable amount of uncertainty about the pandemic’s severity. It might only be-

come obvious in retrospect that the pandemic had been mild.

Short-Term Effects

The most immediate impact of a pandemic would be a surge in demand for medical

services. During a severe pandemic, hospitals, clinics, and doctors’ offices would prob-

ably be overwhelmed, and surveillance (keeping track of where the disease was and

where it was going) would be difficult. Health care workers would be exposed to the

disease, resulting in further strains on the health care system’s capacity, as some work-

ers became sick and others stayed home to care for family members or to avoid be-

coming ill. Care for nonacute health problems would be sharply curtailed.

As the pandemic progressed, international travel would dramatically decline, as peo-

ple avoided avian flu “hotspots” and governments restricted travel. It seems unlikely

that domestic and international air travel would cease completely, but as a point of

reference, at the peak of the SARS outbreak in April 2003, airline passenger arrivals in

Hong Kong had declined by nearly two-thirds relative to their levels in March.

In all likelihood, people would quarantine themselves and their families by staying at

home more. Nonessential activities that required social contact would be sharply cut,

which would lead to significant declines in retail trade. People would avoid public

places, such as shopping malls, community centers, places of worship, and public

transit. Attendance at theaters, sporting events, museums, and restaurants would de-

cline. It seems likely that many schools would close, and even if they did not, atten-

dance would fall dramatically as parents kept their children at home. In either event,

large-scale school closings would lead to a spike in workplace absences because parents

would stay home to care for their children even if they were not sick.

16. See Alan Siu and Richard Wong, “Economic Impact of SARS: The Case of Hong Kong,” Asian

Economic Papers, vol. 3, no. 1 (Winter 2004), p. 76. The authors note that air travel recovered rap-

idly as the epidemic subsided, returning to its pre-epidemic level by August 2003.

The general slowdown in economic activity would reduce gross domestic product

(GDP). Business confidence would be dented, the supply of labor would be restricted

(owing to illness, mortality, and absenteeism spurred by fear of contracting the dis-

ease), supply chains would be strained as transportation systems were disrupted, and

arrears and default rates on consumer and business debt would probably rise some-

what. It seems quite likely that the stock market would fall initially and then rebound

later, as it did in Hong Kong during the SARS episode.

Of course, economic activity would slow, but it would not halt completely. Experi-

ence with such catastrophes as natural disasters and terrorist attacks has demonstrated

the ability of people to cope with and adapt to extremely difficult circumstances.

Moreover, the advances in technology of recent years would allow many companies,

especially those in service industries, to conduct business via electronic communica-

tions, which would permit their employees to work from home. And to the degree

that shipping companies were operating, on-line purchases could offset some of the

decline in retail trade.

The actions of governments could influence the effects of a pandemic on the econ-

omy. Attempts to quarantine people would probably amplify the reductions in trade,

travel, and tourism. However, government actions could also help mitigate economic

impacts. Effective global surveillance and prompt identification of the pandemic

strain by government agencies—along with quarantine and social isolation—could

provide the opportunity for manufacturers to develop a vaccine to lessen the human

and economic costs of a pandemic during its latter stages.

Little information is available for estimating the magnitude of the short-term eco-

nomic impact of a pandemic. CBO was unable to find any estimates of the short-run

economic effects of the three flu pandemics during the 20th century. Consequently, it

based its estimate on three strands of analysis:

B A rough estimate of the supply-side effects from a large share of the labor force’s be-

coming ill;

B A very rough estimate of a pandemic’s effect on demand in individual industries;

and

B A comparison with the impact of the SARS epidemic in Southeast Asia and

Canada.

Effects Under a Severe Pandemic Scenario. To calculate the supply-side impact of a

severe pandemic, CBO combined a rough estimate of the loss of employee work days

with an estimate of average productivity per worker. For most sectors of the economy,

CBO assumed that, on average, 30 percent of the workers in each sector would be-

come ill and of those workers, 2.5 percent would die. Further, CBO assumed that

those who survived would miss three weeks of work, either because they were sick, be-

cause they feared the risk of infection at work, or because they needed to care for fam-

ily or friends. For the farm sector, where the work generally requires less social interac-

tion, CBO assumed that the impact would be milder: one-tenth of workers would be

affected and survivors would miss only a single week of work (the case fatality rate

would be the same). CBO used average productivity per worker, by sector, for 2004 to

compute the impact on GDP of the employment lost to the pandemic. Under the

assumptions detailed above, GDP would be about 2-1/4 percent lower in the year in

which the pandemic occurred, CBO estimates, than it would have been had the pan-

demic not taken place (see the Technical Appendix for a more complete description of

these calculations).

In addition to workers’ absences, many businesses (such as restaurants and movie the-

aters) would probably suffer a falloff in demand because people would be afraid to pa-

tronize them or because the authorities would close them. To calculate those demand-

side effects, CBO examined GDP by industry and assumed different declines in de-

mand for different industries, based loosely on Hong Kong’s experience with SARS.

Those assumed effects were based on judgments about the degree of social interaction

required in different industries and are extremely rough. CBO assumed that a pan-

demic’s effects would be especially severe among industries whose products required

that customers congregate; examples include the entertainment, arts, recreation, lodg-

ing, and restaurant industries. Other industries, including retail trade, were assumed

to suffer a smaller decline in demand, and one industry, health care, was assumed to

experience an increase in demand because of the surge in demand for medical care.

The estimated demand-side impacts sum to about 2 percent of GDP; combining

them with the supply-side impacts implies about a 4-1/4 percent reduction in GDP

in the year of the pandemic.

That scenario suggests that a severe influenza pandemic would have an impact on the

U.S. economy that was slightly larger than the typical recession experienced during

the period since World War II. On average, real (inflation-adjusted) GDP declined by

0.6 percent during the four quarters following each of the 10 business cycle peaks be-

tween 1947 and 2005. Those declines indicate that the average postwar recession low-

ered real GDP by about 4.1 percent, relative to a baseline in which output continued

to grow according to its long-run trend of 3.5 percent. In addition, the estimated ef-

fect on real GDP in the severe scenario exceeds the impact of every postwar recession

except the one following the 1981 peak, which pushed real GDP more than 7 percent

below trend in 1982.

17. This is an estimate of the immediate disruption to the economy caused by the pandemic, not an

estimate of what society would be willing to pay to avoid a pandemic. That amount would be

much larger than the loss of GDP because it would include, among other things, the net present

value of expected future lifetime earnings of those who died from the disease. See Meltzer, Cox,

and Fukuda, “The Economic Impact of Pandemic Influenza in the United States.”

Effects Under a Mild Pandemic Scenario. The economic effects of a mild pandemic

would be much smaller and might not even be distinguishable from the normal ups

and downs of economic activity. To calculate the supply-side effect, CBO assumed

that the attack rate would be 25 percent (except in the farm sector, where it was as-

sumed to be 5 percent), the case fatality rate would be just over 0.1 percent, and the

time out of work would be one-quarter of the duration assumption for the severe sce-

nario (that is, just less than four days absent, on average). Under those assumptions,

GDP would decline by about 1/2 percent (about $70 billion in 2004) as a result of

supply-side factors. For the demand-side effects, CBO assumed that the declines in

each industry would be one-quarter of the declines under the severe scenario, which

amounted to 1/2 percent of GDP (about $60 billion in 2004). In total, the decline in

output amounts to about 1 percent of GDP, relative to what would have happened in

the absence of a pandemic. Compared to the long-run growth trend, a mild influenza

pandemic would cause growth to slow, but would probably not cause real GDP to fall

(or cause a recession).

Comparison with SARS. Past studies of the epidemic of severe acute respiratory syn-

drome that affected Southeast Asia, China, and Canada provide a point of compari-

son for CBO’s estimates. The scale of the SARS epidemic—more than 8,000 cases

worldwide and nearly 800 deaths—was much smaller than the pandemics considered

in the scenarios CBO developed for this analysis. Nevertheless, that experience is ap-

plicable to a potential avian flu pandemic because people’s reaction to the SARS out-

break could be considered similar to what might be expected in a flu pandemic. A

study by Jong-Wha Lee and Warwick McKibbin used a global macroeconomic model

to estimate the impact of the SARS epidemic on GDP in 2003. Lee and McKibbin es-

timate that the decline in GDP was largest in Hong Kong, where output was 2.6 per-

cent lower in 2003 than it would have been without a pandemic. Hong Kong was fol-

lowed by China, according to the authors, with a decline of 1.1 percent of GDP

relative to a base case without a pandemic, and by Taiwan and Singapore, with de-

clines of about 0.5 percent.

However, a separate study by Alan Siu and Richard Wong, which focused exclusively

on Hong Kong, concluded that the SARS epidemic had had only a mild impact on

output in 2003.

Although they do not present their own estimates, they conclude

that private-sector forecasts predicting that SARS would knock 1.2 percentage points

off the growth of GDP proved to be too pessimistic. That the disease’s impact was

mild was not because aggregate demand did not fall. The authors document declines

in travel, tourism, and retail trade that were severe: year-over-year declines in retail

18. See Jong-Wha Lee and Warwick McKibbin, Globalization and Disease: The Case of SARS, Working

Paper No. 2003/16 (Australian National University, August 2003).

19. See Siu and Wong, “Economic Impact of SARS.” Hong Kong accounted for about 22 percent

(about 1,800) of the total number of SARS cases worldwide and 40 percent (about 300) of the

deaths.

trade of 15 percent in April 2003, a decline of 63 percent in total visitors arriving in

Hong Kong, and sharp declines in restaurant spending.

The authors argue that

SARS’s economic impacts were concentrated in the sectors that were most directly af-

fected by the epidemic and were relatively short-lived. Moreover, the epidemic caused

no major disruption to international trade nor to Hong Kong’s manufacturing base in

the Pearl River Delta region of China.

An August 2005 study by the investment firm BMO Nesbitt Burns also examined the

economic impact of the SARS. It describes the reaction to the epidemic in Toronto

(which included sharp drops in travel, retail trade, and tourism) and documents the

strains on the health care system. Citing a study by the Bank of Canada, the study es-

timates that the crisis cut GDP in the second quarter of 2003 by 0.6 percent, which

implies that economic activity in Toronto declined by about 3 percent in that quarter

and roughly 0.75 percent for the year.

The estimates of the economic effects of the SARS epidemic provide a useful bench-

mark for CBO’s estimates of the impact of an influenza pandemic. The SARS esti-

mates also show how difficult it is to estimate the economic effects of an epidemic.

Even though all of the studies were performed after the fact (using actual data), their

results vary widely. Nevertheless, the estimates suggest that the impact of SARS on

Hong Kong’s economy (which is where the effects were concentrated) was roughly

similar to a mild business-cycle recession, although the studies differ sharply about the

degree of the slowdown. CBO’s estimates of the effects of the two influenza pandem-

ics are roughly comparable. The effect on GDP in the severe scenario is larger than

any of the estimates of SARS’ effects, whereas the estimated effects from the mild sce-

nario are in the middle of the range of SARS estimates.

One could argue that what happened with SARS is not relevant for an analysis of the

effects of an influenza pandemic in the United States because people with influenza,

unlike those with SARS, are contagious before the onset of case-defining symptoms.

As a result, control measures that are based on case identification (which worked well

for SARS) will be ineffective for the flu, and a much larger proportion of the popula-

tion is likely to be infected.

Consequently, an influenza pandemic might be ex-

pected to have a larger impact on the economy than did SARS.

There are also reasons, however, to think that the SARS epidemic is relevant for ana-

lyzing the effects of a flu pandemic. First, Hong Kong is a densely populated city; the

impact of the flu in cities like New York and Boston might exceed that of SARS in

Hong Kong by a wide margin, but the flu’s impact on the overall economy might be

20. Tourism makes up a much larger share of Hong Kong’s economy—about 6 percent—than it does

of the United States’ economy, and tourists contribute more to retail sales there than they do in the

United States.

21. See Christina Mills, James Robbins, and Marc Lipsitch, “Transmissibility of 1918 Pandemic Influ-

enza,” Nature, vol. 432 (December 16, 2004), pp. 904-906.

considerably less. Second, the economic effects of the SARS outbreak appear to have

been disproportionate to the virulence of the disease. The disruption to travel, trans-

port, and tourism was substantial in Hong Kong and Toronto, as people shunned

public places and public transportation, thus limiting the scope for even larger de-

clines in economic activity in response to a deadlier disease.

Long-Term Effects

The most important long-term impact of a pandemic is the reduction that would per-

sist in the population and in the labor force after overall demand in the economy re-

turned to normal. The effects of that drop in the population would depend, in part,

on the characteristics of the outbreak. If, for example, mortality was concentrated

among the very young and the very old, then a pandemic would have relatively small

effects on the subsequent growth of GDP. By contrast, if the disease struck workers

who were in their prime working years more heavily, then the effects on GDP growth

during the years following the pandemic would be more significant. However, predic-

tions of the size and direction of those effects are ambiguous. Under standard assump-

tions, for example, a one-time reduction in the labor force would raise the ratio of

capital to labor and lower the rate of return to capital, thus slowing the pace of capital

accumulation and GDP growth for many years. However, under other types of analy-

ses that include the influence of human capital, reduction in the labor supply would

encourage investment in human capital, which would tend to speed the growth of per

capita output during the transition period, when the economy was recovering from

the shock.

How big a reduction in the U.S. labor force would a severe pandemic produce? In

2004, the labor force totaled 147.4 million people. Under the assumption of an attack

rate of 30 percent and a case fatality rate of 2.5 percent—the same assumptions ap-

plied to the population as a whole—a severe pandemic would cause the deaths of

more than 1 million labor force participants, or about 0.75 percent of the labor force.

Since growth in the labor force averaged 1.6 percent during the 1948-2005 period,

losing 0.75 percent of the labor force would be equivalent to a pause of one-half year

in the growth of the work force.

Under the assumptions for infection and mortality

associated with the mild-pandemic scenario—an attack rate of 25 percent and a fatal-

ity rate of just over 0.1 percent—the number of workers killed would be 50,000, or

0.03 percent of the labor force.

There is little evidence available to use to determine which theoretical prediction best

describes the long-term impact of an influenza pandemic. One study that analyzed

whether the decline in population associated with the plague in Europe during the

14th century caused an increase in wages (and, therefore, in per capita output) re-

22. A decline of that magnitude is not particularly large relative to the normal yearly variation in labor

force growth. For example, the standard deviation (a measure of the variability of a data series) of

labor force growth during the 1948-2004 period is 0.8 percentage points, which indicates that

labor force growth was within 0.8 percentage points of its mean value (of 1.6 percent) about two-

thirds of the time during the historical sample period.

ported inconclusive evidence.

That study also looked at the effect of the 1918 in-

fluenza pandemic on agricultural production across 13 Indian provinces and found no

relationship between changes in population and agricultural output. By contrast, in a

study that analyzed the effects of the 1918 flu pandemic on economic growth using

data from U.S. states, the authors found that after controlling for other variables—

such as population density, education levels, share of foreign-born population, and

climate—states that were harder hit by the epidemic experienced faster per capita

growth during the 1919-1930 period.

The authors caution, however, that their re-

sults may be picking up a transitional effect—a return to trend growth—rather than

an increase in steady-state growth. Moreover, because the study looks at growth rates

among the states, it does not address the question of whether the growth of overall

GDP would be faster or slower.

Finally, there is evidence that a potential influenza pandemic could have an impact

over the very long term, lowering the incomes and socioeconomic status and harming

the health of children who were born after the pandemic and were in utero during its

height.

Current Policies and Options for the Future

Preparedness for a flu pandemic lies within the nation’s overall capacity to address

public health emergencies. Growing concern about a flu pandemic occurs at a time

when the public health system overall and the parts of it directly associated with influ-

enza are under increased scrutiny, and when heightened concerns about terrorism

have been the primary driver of increased public health funding.

Options for addressing potential pandemics in the near and long term fall into four

broad categories: detecting and controlling viruses at their source, developing and us-

ing vaccines to prevent diseases, developing and using treatments with antiviral drugs

and other medications, and building the capacity of the health care system (facilities,

equipment, and people) to deliver care. Decisions made in the near term may alter the

availability of vaccines and treatment options in the longer term. In making decisions

about how to proceed, there are competing risks. The risk of inaction is that a pan-

demic will occur that could have been prevented or mitigated. The risk of action is

23. See David Bloom and Ajay Mahal, Does the AIDS Epidemic Really Threaten Economic Growth?

Working Paper No. 5148 (Cambridge, Mass.: National Bureau of Economic Research, June

1995). That paper was later published under the same title in the Journal of Econometrics, vol. 77,

no. 1 (1997). See also Bloom and Mahal, “AIDS, Flu, and the Black Death: Impacts on Economic

Growth and Well Being,” in David Bloom and Peter Godwin, eds., The Economics of HIV and

AIDS: The Case of South and South East Asia (Delhi: Oxford University Press, 1997).

24. See Elizabeth Brainerd and Mark Siegler, The Economic Effects of the 1918 Influenza Epidemic, Dis-

cussion Paper No. 3791 (London: Centre for Economic Policy Research, February 2003).

25. See Douglas Almond, “Is the 1918 Influenza Pandemic Over? Long-term Effects of In Utero Influ-

enza Exposure in the Post-1940 U.S. Population” (mimeo, Columbia University, July 2005).

that it is costly, diverts resources from other priorities, and could be damaging by it-

self, as was the case in the reaction to the swine flu scare of 1976.

Context

Rising concerns in the 1990s about the spread of disease as a result of international

commerce and travel, antibiotic resistance, biological and chemical terrorism, and

new and evolving pathogens led to an increased focus on improving the public health

system. Federal funding, initially targeted toward efforts to stockpile medicines and

improve laboratory capacity, increased substantially in the face of concern about the

medical consequences of terrorism after the terrorist attacks of September 11, 2001,

and the anthrax attacks in that same year. Between 2001 and 2005, policymakers ap-

propriated almost $30 billion for activities to address the nation’s preparedness against

infectious diseases; improve cooperation between federal, state, and local authorities

in the event of a public health emergency; upgrade the nation’s laboratory capacity

and expand the ability of local governments to deal with large numbers of people

needing health care. Over half of that funding was for expanding research on vaccines

and therapies against infectious diseases. Other funds aim to encourage private firms

to develop new vaccines and other remedies to respond to biological terrorism and

improve the national stockpile of and distribution strategy for remedies in the event of

such an attack.

Countering bioterrorism has been the main priority of those efforts and about $15

billion was appropriated for those purposes between 2001 and 2005. Many of the

preparedness activities supported by that spending contribute to stronger health care

systems and improved processes for combating epidemics of infectious diseases, in-

cluding pandemic influenza. But concerns remain about how to specifically address

an influenza pandemic. First, are concerns about the health system’s ability to care for

a large number of people in all parts of the country who are sick at the same time. Sec-

ond, the declining number of suppliers of flu vaccine and the vaccine production

problems that occurred in the 2001-2002 and 2004-2005 flu seasons highlight con-

cerns about the national capacity to produce vaccines that would mitigate the damage

a pandemic outbreak might cause. Concerns about a pandemic flu may have increased

the urgency to resolve problems that rose to prominence last year in relation to the

pace of technical improvement in the vaccine industry—an issue of long-standing

concern for public health analysts—and may have increased the willingness to pay for

solutions. Third, heightening those worries are more-recent developments indicating

that the worldwide demand for antiviral drugs that might blunt the impact of an out-

break far exceeds current production capabilities.

Over the past several months, both the Administration and Members of Congress

have proposed appropriating between $4 billion and $8 billion specifically for pro-

grams to respond to the prospect of a flu pandemic. Those proposals raise a number

of questions. For example, how would costs compare with net expected benefits? Has

the right balance been struck between spending for health system response and vac-

cines and drugs? How can significant new spending for a flu pandemic be best incor-

porated in the larger public health effort? What lessons have been learned thus far,

about how government policies can improve and guarantee the supply of vaccines and

antiviral drugs? And how can effective policies be formulated in the presence of great

uncertainty?

This section summarizes the Administration’s recently released plan and some Con-

gressional alternatives. It also lists and examines specific policies now in effect and op-

tions being considered whose intention is to limit the harm of a pandemic flu. Those

proposals and policies may be revised as scientists and policymakers learn more about

the virus and possible countermeasures. In keeping with CBO’s mandate to provide

objective, nonpartisan analysis, this paper makes no recommendations.

The discussion of options and policies addresses both measures that are directed to-

ward a near-term outbreak and measures that increase the nation’s capability to deal

with a pandemic several years into the future. Uncertainty is an ever-present concern,

prompting such questions as, Will the H5N1 strain be the pandemic strain? Will the

vaccines currently being stockpiled prove effective? Likewise, will antiviral drugs prove

useful? And can the public health system cope with the surge in demand that an out-

break would bring? Not surprisingly, most of the policies now in effect and those be-

ing contemplated can be analyzed only by making heroic assumptions about how

such uncertainty is resolved. Some small comfort may be found in the fact that in sev-

eral instances, the issues being raised are not unique to policy regarding flu pandem-

ics. Such issues include, for example, how incentives to produce new antiviral drugs

can be preserved if the patent rights of the current producers of those drugs are not re-

spected or how liability for the harm a vaccine might cause can be best distributed

among producers, consumers, and taxpayers.

The Administration’s Proposal and Congressional Alternatives

In November 2005, the President requested a total of $7.1 billion in emergency fund-

ing for influenza pandemic preparedness. About 95 percent of that amount was to be

spent on producing and stockpiling vaccines, antiviral drugs, and other medical sup-

plies and on enhancing surveillance systems both internationally and domestically, as

described in the Department of Health and Human Services (HHS) Pandemic Influ-

enza Plan (hereafter referred to as the HHS plan).

26. Sources for CBO’s discussion of the HHS plan include Michael O. Leavitt, Secretary of the

Department of Health and Human Services, Pandemic Influenza, statement before the Subcom-

mittee on Labor, Health and Human Services, and Education of the Senate Committee on Appro-

priations, November 2, 2005, available at http://appropriations.senate.gov/hearmarkups/

Leavitt.htm; a speech by President Bush at the National Institutes of Health on November 1,

2005, “President Outlines Pandemic Influenza Preparations and Response, available at

www.whitehouse.gov/news/releases/2005/11/print/20051101-1.html; “White House Fact Sheet:

Safeguarding America Against Pandemic Influenza,” November 1, 2005, available at http://www.

whitehouse.gov/news/releases/2005/11/20051101.html; and Department of Health and Human

Services, HHS Pandemic Influenza Plan (November 2005), available at http://www.hhs.gov/

pandemicflu/plan/.

Several other proposals that have been introduced in the Congress also address pre-

paredness for an influenza pandemic. Although they share the same general goals,

they differ from each other and from the Administration’s proposal in levels of fund-

ing, relative emphasis on vaccine stockpiling versus other activities, the extent to

which vaccine manufacturers and providers would be protected from liability lawsuits,

and the share of the costs that state and local governments would be required to con-

tribute. Several proposals would provide greater funding for antiviral drugs than for

vaccines, whereas others would allot greater funding for public health preparedness.

Recent proposals to expand Project Bioshield (discussed later) would fold incentives

for the development of pandemic flu vaccines and drugs into broader efforts to pro-

vide incentives for developing drugs to treat effects arising from bioterrorist attacks.

Still other proposals would focus narrowly on liability protections for manufacturers

and providers of vaccines and other flu countermeasures.

Under the HHS plan, $4.7 billion would go toward investments in creating produc-

tion capacity for pandemic influenza vaccine and a stockpile of enough vaccine

against each circulating influenza virus with pandemic potential to provide a course of

inoculation for 20 million people by 2009.

The ultimate goal is to have a “surge”

capacity sufficient to produce enough vaccine for the entire U.S. population (almost

300 million people) within six months of a pandemic outbreak.

The Administration argues that sufficient capacity cannot be achieved from egg-based

production alone; its proposal is based on only 20 percent of surge capacity coming

from that source and 80 percent coming from new cell-based vaccine-manufacturing

facilities. To achieve that goal, HHS would spend $2.8 billion to finance the establish-

ment of new cell-based vaccine manufacturing facilities that could open by 2010. The

department would also finance the retrofitting of existing domestic egg-based manu-

facturing facilities.

Following are several other features of the HHS plan:

B HHS would work with industry and academia to support development of dose-

sparing technologies (that is, technologies that stimulate a strong immune response

using less antigen, the raw material of vaccines) and would invest in research to de-

velop a universal vaccine that would work against all strains of the flu.

27. A course is two doses of vaccine, each with 90 μg (micrograms) of antigen.

B The Administration proposes liability protections for vaccine manufacturers and

providers on the grounds that the threat of litigation is an obstacle to the develop-

ment and production of new vaccines.

B Under the HHS plan, $1.4 billion would go toward purchasing antiviral drugs and

accelerating the development of new antiviral drugs. About $1 billion of those

funds would be used to purchase 50 million courses of treatment of antiviral drugs

and to subsidize the states’ purchase of an additional 31 million courses under fed-

eral contracts.

B HHS would spend $555 million to expand domestic and international surveil-

lance, strengthen public health infrastructure, and communicate with the public

about the risks of an influenza pandemic. Because state and local governments

would be largely responsible for distributing vaccines, antiviral drugs, and other

medical supplies, the plan designates $100 million of the $555 million for state

and local pandemic preparedness efforts.

Near-Term Options

Over the next several months, the options available to prevent a possible pandemic or

to diminish its effects are limited. The United States could provide technical and fi-

nancial support for national and multilateral public health organizations that monitor

potential outbreaks and control them at their source by culling flocks of infected

birds. On the domestic front, were an outbreak to occur, limited production capacity

of vaccines and antiviral drugs would be used strategically to slow a potential pan-

demic and limit the impact on the health care system. Equally important, policy deci-

sions in several of those areas would have the potential to improve the longer-term na-

tional capability to cope with an outbreak.

28. HHS would subsidize 25 percent of the cost of the antiviral drugs purchased by state

governments.

Source Control and Surveillance. The longer the H5N1 virus remains transmissible

only to humans who are in close contact with infected birds, the more likely it is that

vaccines, antiviral drugs, and public health preparedness in general will be useful in

limiting the effects of a possible pandemic. In the event that the virus becomes easily

transmissible from person to person, the sooner that fact is known, the more time

there will be to gather and deploy the available public health resources.

Current policy recognizes the importance of source control and surveillance. The

United States participates in and contributes to the World Health Organization’s

(WHO’s) and other international agencies’ efforts to contain the H5N1 epidemic

among poultry flocks (by educating farmers in proper bird handling, for example) and

to detect new outbreaks. For example, the U.S. Agency for International Develop-

ment is spending $13.7 million for avian influenza control, prevention, and prepared-

ness in Asia, complementing $6 million in spending on international surveillance by

the Centers for Disease Control and Prevention (CDC).

As the H5N1 epidemic has progressed in bird populations, source control has become

an increasingly important tactic to delay the onset of a potential H5N1-based pan-

demic in humans. Although the social benefits of identifying and culling infected

flocks are clear, the incentives for individual farmers to actively cooperate are less

clear-cut. Providing compensation to farmers who are vigilant and proactive could

improve the likelihood that source control will be successful. Accordingly, ensuring

that poorer countries have the funds necessary to provide such compensation as well

as undertake surveillance and educational activities is in the interest of the United

States and other developed countries. A November 2005 report prepared by the

United Nations Food and Agriculture Organization questions whether the funds cur-

rently committed on a bilateral and multilateral basis are sufficient to implement the

best strategy to control avian flu in the animal population.

Although support for countries and individuals coping with avian influenza in their

domesticated bird populations may be desirable, certain risks are evident. Nations or

farmers who expect outside help to pay for a problem that affects them as well as ev-

eryone else will have diminished incentives to spend resources of their own to identify

infected flocks and cull them. Thus, the payoff from more aid for those purposes is

likely to be offset in part by recipients’ weaker incentives to take action because they

are expecting someone else to pay.

29. U.N. Food and Agriculture Organization, A Global Strategy for the Progressive Control of Highly

Pathogenic Avian Influenza (HPAI) (Rome: U.N. Food and Agriculture Organization, November

2005), p. 13-14.

Another option for source control in the near term is vaccinating the domestic bird

flocks of nations where avian influenza has been detected. The Chinese government is

undertaking such an effort, and some recent research suggests that vaccinating chick-

ens may be an effective means of source control.

Again, it may be in the interests of

developed nations to subsidize vaccination programs in poor countries. But inefficient

vaccination giving only partial immunity to inoculated flocks could do more harm

than good by making it more difficult to detect a lethal strain of avian flu in domesti-

cated flocks.

Vaccines. Currently, no licensed human vaccine is available for the strain of the avian

influenza virus now circulating in Southeast Asia. The National Institutes of Health

(NIH) has contracted with private manufacturers to produce developmental vaccines

against the H5N1 strain and has tested those vaccines in healthy adults; it plans more

testing among children and the elderly. This past fall, the U.S. government awarded

$162.5 million in contracts for additional vaccine development to two firms: Chiron

and sanofi pasteur.

Very recent reports indicate that the government intends to have

a stock of vaccines sufficient to immunize 4 million people against the H5N1 virus by

February 2006.

If the H5N1 flu strain triggered a pandemic in the next several months, the Food and

Drug Administration (FDA) would probably request that producers switch from

manufacturing seasonal flu vaccine to producing vaccine effective against the pan-

demic strain. Currently, the only manufacturer of flu vaccine located in the United

States is sanofi pasteur, which can produce 60 million doses of seasonal vaccine about

every six months. Each dose of seasonal vaccine contains 45 μg (micrograms) of anti-

gen, the raw material from which vaccines are made. The best available science indi-

cates that two doses of vaccine, each requiring 90 μg of antigen, are necessary to pro-

tect against the H5N1 strain.

Thus, the raw material that currently can be used to

inoculate 60 million people against the seasonal flu would provide immunity to only

30. J.A. van der Goot and others, “Quantification of the Effect of Vaccination on Transmission of

Avian Influenza (H7N7) in Chickens,” Proceedings of the National Academy of Sciences, vol. 10

(December 5, 2005).

31. Chiron’s contract was awarded in October and is for $62.5 million; sanofi pasteur’s contract was

awarded in September and is for $100 million.

32. Statement of Michael Leavitt, November 2, 2005, p. 5.

15 million people every six months during a pandemic caused by the H5N1 strain.

Optimal vaccination of the United States’ 300 million people would require many

times that amount. That shortage of capacity results in part from problems in the

market for pandemic influenza vaccine, many of which are present in the seasonal in-

fluenza market as well (see Box 1).

Some experts believe that it is overly optimistic to assume that antigen yields from us-

ing the capacity available to produce seasonal influenza vaccine would be realized in

producing a vaccine effective against the H5N1 virus. Because H5N1 is lethal to

birds, it both threatens the supply of chickens able to provide the necessary eggs and

increases the number of eggs required for each dose. The pharmaceutical manufac-

turer GlaxoSmithKline has conducted tests that indicate that the yield for a pandemic

strain could be 20 percent to 50 percent lower than it is for each strain of seasonal vac-

cine—which would reduce the number of people who could be vaccinated propor-

tionately.

In the event of a near-term flu pandemic, the short supply of vaccine and the likeli-

hood of extraordinary demand for it could require the public health system to take a

forceful role in distributing and administering the limited supplies of vaccine. The

HHS plan lists priority groups as recommended by its Advisory Committee on Im-

munization Practices (ACIP) and its National Vaccine Advisory Committee (NVAC).

The plan states that the “primary goal of a pandemic response considered was to de-

crease health impacts including severe morbidity and death; secondary pandemic re-

sponse goals included minimizing societal and economic impacts.”

ACIP and

NVAC recommendations give priority to medical workers (8 million to 9 million

people) and people critical to the manufacturing of vaccines and antiviral drugs

(40,000 workers). The recommendations then give priority to the elderly with

33. Experience to date indicates that the H5N1 strain presents special problems for vaccine develop-

ment. For example, laboratories have been unable to prepare the candidate H5N1 seed strain in

the same manner as they prepare the candidate seasonal influenza A seed strains. Those seasonal

seed strains are prepared by genetic reassortment using a strain that improves growth performance.

If the candidate H5N1 seed strain was prepared in that manner, it would still be capable of caus-

ing severe disease and therefore would not be suitable for manufacturing. Instead, the candidate

H5N1 seed strain must be prepared by reverse genetics to make it incapable of causing disease and

suitable for manufacturing. MedImmune holds the patent rights for a reverse-genetics process use-

ful in vaccine manufacturing. It has allowed manufacturers to produce pandemic influenza vac-

cines from seed strains prepared by this process for research purposes but not for distribution.

However, according to officials at NIH, MedImmune has agreed to license the technology to vac-

cine manufacturers in return for a licensing fee. Resolving such license issues now might speed up

the production of pandemic flu vaccine in the event of an outbreak.

34. Andrew Jack, “Fighting Flu: Vaccine Makers Roll Up Their Sleeves,” Financial Times, October 30,

2005.

35. Department of Health and Human Services, HHS Pandemic Influenza Plan, Part 1, Appendix D,

p. D-10.

Box 1.

Problems in the Market for Influenza Vaccine

Planning, production, and distribution of influenza vaccine follow an annual

cycle. The Centers for Disease Control and Prevention (CDC) runs a domestic

flu surveillance program and works with the World Health Organization

(WHO) and other groups to monitor flu internationally. Each year, those ef-

forts culminate in a decision about which strains of flu should be included in

that year’s seasonal vaccine. The WHO Collaborating Centers for Influenza,

the U.S. Food and Drug Administration (FDA), and other agencies’ laborato-

ries prepare candidate high-growth seed strains and provide them to manufac-

turers licensed by the FDA. The manufacturers report back to the WHO Col-

laborating Centers and the FDA on the candidates’ suitability for vaccine

production. Sometime between February and March, FDA announces the se-

lected strains to be included in the vaccine for the upcoming flu season. Manu-

facturers must then produce enough vaccine for the market by October to No-

vember (the optimal time for vaccinations), using a process that can take six

months or more and is based on growing the viruses in fertilized chicken eggs.

Experience in both the 2000-2001 and 2004-2005 flu seasons makes it clear

that much can and does disrupt that annual cycle.

The following are among

the problems in the flu vaccine market:

B The lengthy egg-based manufacturing process means that production can-

not be scaled up quickly if flu vaccine demand is higher than predicted.

Also, demand is hard to predict and can depend on such things as the tim-

ing and media coverage of the current flu season and the severity of previous

flu seasons. Demand for vaccine for a pandemic flu is even more variable,

ranging from none in most years to perhaps 10 times more (by quantity of

antigens) in a pandemic year than the average flu vaccine demand.

1. This description of the annual seasonal-flu cycle was drawn from Catherine Gerdil, “The

Annual Production Cycle for Influenza Vaccine,” Vaccine, vol. 21 (2003), and personal

communication to the Congressional Budget Office from staff of the National Institutes of

Health.

2. Government Accountability Office, Flu Vaccine: Supply Problems Heighten Need to

Ensure Access for High-Risk People, GAO-01-624 (May 2001), and Influenza Vaccine:

Shortages in 2004-05 Season Underscore Need for Better Preparation, GAO-05-984

(September 2005).

Box 1.

Continued

B Besides being highly variable, demand for flu vaccinations is likely to be

smaller than the socially optimal level. Flu vaccine helps not only the person

vaccinated but also people with whom that person comes in contact, be-

cause each vaccinated person adds to “herd immunity” and decreases the

possibility of transmission. Individuals’ decisions about the desirability of

getting a flu vaccination may not take into account that additional social

benefit, however. People may also choose not to get vaccinated if they be-

lieve that everyone with whom they come in contact will be vaccinated.

B Flu vaccine cannot be stored from one flu season to the next because flu

strains change from year to year. Consequently, if manufacturers supply

more vaccine than demanded, the excess vaccine is destroyed.

B The manufacturing process is prone to contamination, which upon inspec-

tion will cause vaccine to be withheld from the market.

B Although the government accounts for less than 20 percent of the market

for flu vaccine, the CDC’s role in allocating vaccine during shortages dis-

courages the establishment of reserve capacity. In the event of shortages, the

CDC recommends that vaccine be given only to people in priority groups in

order to decrease the health effects associated with the flu, including severe

morbidity and death. If, instead, market incentives were allowed to deter-

mine the allocation of vaccine during periods of scarcity, manufacturers

would have a greater incentive to increase their production capacity. In that

case, however, people in priority groups who were unwilling to pay enough

might not get vaccinated.

Because of those risky market conditions, few manufacturers of flu vaccine

serve the U.S. market, which makes that market vulnerable to supply disrup-

tions.

conditions that put them at high risk for hospitalization and death if they become in-

fected with influenza (18.2 million people) and to people ages 6 months to 64 years

with two or more conditions that put them at high risk for hospitalization and death

if they become infected with influenza (6.9 million people).

Some critics argue that younger and healthier individuals should be given priority

over the elderly and over people with conditions that put them at high risk for hospi-

talization and death if they become infected with influenza. Such critics argue that

younger and healthier individuals are more mobile than older, less healthy people and

therefore are more likely to spread the flu to others. Healthier individuals are also

more likely to play critical roles in supply chains for food, water, electricity, and the

like.

Another factor arguing for giving priority to younger people is that the sea-

sonal flu vaccine produces a weaker immune response in the elderly, making it less ef-

fective in flu prevention. Finally, a potential pandemic flu could have characteristics

similar to those of the flu of 1918, which caused a disproportionately high number of

deaths among the young and healthy.

The discussion above assumes that two doses of 90 μg each is the course of treatment

administered to each person. An alternative strategy would be to reduce that course

and inoculate a much larger number of people. What level of immunity might be pro-

vided from a diluted course is unknown—although initial studies by NIH found that

diluted courses did not achieve the desired immune response. The risk of the dilution

strategy is that sufficient immunity will not be provided to key priority groups whose

immunization is essential to mitigating the effects of a pandemic—groups such as

medical workers and people critical in manufacturing vaccines and antiviral drugs.

Even if supplies of vaccine were adequate, having the government conduct a mass im-

munization campaign for a potential epidemic is an enterprise not without challenges

and risk. The swine flu response in 1976 serves as a cautionary tale for mass vaccina-

tion programs and raises the question of whether the risks of such a program merit ac-

tion before an outbreak actually occurs. Concerned that the infection of four soldiers

at Fort Dix with the swine flu presaged a pandemic akin to the 1918 Spanish flu,

health officials pressed for a national vaccination campaign. Although some health of-

ficials expressed doubts about the likelihood of an epidemic, the government initiated

a mass inoculation program and requested that manufacturers produce 150 million

new doses to vaccinate the entire U.S. population. After hundreds of people receiving

the vaccine came down with Guillain-Barre syndrome (GBS), a rare neurological dis-

ease thought to be caused by the swine flu vaccine, the government terminated the

36. For the remaining priority groups, see the HHS plan (Appendix D, p. D-13, Table D-1).

37. Tyler Cowen, Avian Flu: What Should Be Done, Working Paper Series (Mercatus Center, George

Mason University, November 11, 2005), p. 11.

campaign and indemnified manufacturers, ultimately paying out $93 million in

claims.

The epidemic never occurred.

More recently, efforts to vaccinate several million health care workers against smallpox

fell short as a result of doubts about the vaccine’s safety and the low probability of a

smallpox attack. The government spent about $500 million to build a stockpile of

vaccine against smallpox for the entire population and began an immunization pro-

gram that would prioritize health care workers. However, the vaccine caused an unex-

pected number of injuries and deaths, and only about 40,000 individuals have been

vaccinated to date.

Antiviral Drugs. A key part of the current strategy for preparing for a flu pandemic is

the purchase and stockpiling of antiviral drugs to prevent infection in people exposed

to the virus and to treat illness. Should an outbreak occur, antiviral drugs would be a

key factor in treating the sick and halting further spread of the flu. Other drugs, espe-

cially antibiotics, would also be necessary to treat the sick.

Only two antiviral drugs, oseltamivir and zanamivir, have shown promise in treating

avian influenza. Oseltamivir is licensed as Tamiflu and manufactured under patent by

the Swiss firm Roche Laboratories in tablet form. A treatment course for Tamiflu in-

cludes 10 pills, taken over five days. Zanamivir is licensed as Relenza and manufac-

tured by GlaxoSmithKline. Relenza is administered by oral inhalation and is not rec-

ommended for people with chronic lung problems.

The FDA has approved both

antiviral drugs for treating influenza; Tamiflu has also been approved to prevent influ-

enza infection.

Because antiviral agents can be stored—Tamiflu and Relenza have shelf-lives of three

and five years, respectively—developing a stockpile has advantages as part of a strategy

for a flu pandemic. Several countries, including Japan, the United Kingdom, France,

Norway, the Netherlands, and New Zealand, are pursuing a stockpiling approach. By

the end of 2005, more than 4 million treatment courses of Tamiflu and 84,000 treat-

ment courses of Relenza will be in the U.S. Strategic National Stockpile.

The

United States has ordered 12 million more treatment courses of Tamiflu from Roche,

but the company’s production capacity is committed to filling previous contracts for

the next two years. Even if the order could be filled, that amount would be enough to

cover only about 5 percent of the population, far short of WHO’s recommendation

that governments have enough antiviral treatments on hand for at least 25 percent of

their population.

38. Sarah Lister, Pandemic Influenza: Domestic Preparedness Efforts, CRS Report for Congress

RL33145 (Congressional Research Service, November 10, 2005), p. 28.

39. The Pill Book, 11th ed. (New York: Bantam Books, 2004), p. 1169.

40. Department of Health and Human Services, Pandemic Influenza Plan, p. F-39.

Securing antiviral drugs, however, is not simply a matter of funding because there are

currently shortages of them. The demand for Tamiflu from governments and private

entities exceeds Roche’s production capacity, and the firm temporarily suspended

shipments of the drug to the United States in late October to prevent hoarding and

ensure adequate supplies to treat seasonal flu cases this winter. In the event of an influ-

enza pandemic in the immediate future, sufficient quantities would not be available to

treat the sick. That situation may improve, however, since Roche will soon begin to

manufacture Tamiflu in the United States (in six new FDA-approved facilities) and

claims that the firm will increase 10-fold over 2003 levels.

Even if the supply of antiviral drugs were adequate, questions would remain about re-

lying too heavily on them for protection. Many of those who died from H5N1 influ-

enza were given Tamiflu, although the drug may have been administered too late or in

too small a dose to be effective.

(Tamiflu needs to be taken within two days of the

initial flu symptoms, when many people may not yet be aware that they have the flu.)

Some research in mice has suggested that Tamiflu is less effective against recent strains

of H5N1 than it was against the 1997 strain.

Additionally, the nature of the treat-

ment (10 pills over five days) might make it difficult for overburdened medical facili-

ties to provide Tamiflu and for patients to comply with the dosage regimen.

An additional concern about antiviral drugs is the location of stockpiles and the distri-

bution of drugs, should an outbreak occur. That concern overlaps with concerns

about health care system readiness, discussed below. Without an adequate plan that

accounts for limited supplies and the predominantly local character of health care de-

livery, a national stockpiling plan will not be effective. Federal, state, and local efforts

to devise such a plan are ongoing.

Health Care System Readiness. If a pandemic occurred in the next several months, the

United States would have to rely primarily on the existing public and private health

system (and its current capacity for hospitalization, antibiotics, and ventilators) to

treat infected people. The effectiveness of that response would depend not only on the

federal government but also on the state and local health authorities and the private

sector, which traditionally play large roles in immunization, laboratory services, and

deciding on and implementing public health actions. Health system response, while

relatively more important than other options were an outbreak to occur in the near

term, would also remain critical for the foreseeable future, in that the effectiveness of

antiviral drugs is uncertain and vaccines, even if produced in mass quantities in the

future, would not be available for at least six months after an outbreak. Limiting the

41. “Roche Plans Big Increase in Tamiflu Production,” New York Times, November 10, 2005; and

“Run on Drug for Avian Flu Has Physicians Worried,” Washington Post, October 22, 2005.

42. Garrett, “The Next Pandemic?”

43. “Mouse Studies of Oseltamivir Show Promise Against H5N1 Influenza Virus,” NIH News, Mon-

day, July 18, 2005, available at www.nih.gov/news/pr/jul2005/niaid-18.htm.

harm of a pandemic flu outbreak may also depend on social distance measures—for

example, closing schools and shopping malls.

Across the nation, pandemic influenza would bring people to physicians’ offices and

health care facilities, which could quickly face overcrowding. Higher-end estimates of

the number of sick individuals treated by hospitals range from 5 million to 10 mil-

lion.

Those figures, even when converted to a bed-day basis, would far exceed the

number of staffed hospital beds. Currently, the United States has approximately

970,000 staffed hospital beds and 100,000 ventilators, with three-quarters of them in

use on any given day. As a result, shortages could occur in critical areas such as venti-

lators, critical care beds, and drugs to treat secondary infections. The ability of facili-

ties to maintain strict infection control would be challenged. Pandemic flu would be

widespread and would restrict the flexibility to shift resources to other communities.

Surge capacity is particularly a concern for the services of health care personnel, espe-

cially nurses, epidemiologists, and laboratory technicians. In many regions of the

country, surge capacity would be further limited by the likelihood that health care

personnel might themselves be sick or be called upon to care for sick family members

at home.

Communities and health care facilities may have to look to other facilities to hold the

sick and adopt diversion strategies for non emergency patients. Additional hospital

bed capacity may be created by setting up field hospitals and using auxiliary sites such

as shelters, schools, religious facilities, nursing homes, hotels and day care centers.

Wide variations exist among communities in the ability of local officials to cope with

an outbreak, with smaller jurisdictions likely at a disadvantage. The suggestion has

been made that government facilities and health care personnel could provide a signif-

icant surge capacity, but as Box 2 discusses, the medical capacity of the Departments

of Defense, Veterans Affairs, and Homeland Security is also limited. Policymakers

may need to adopt a strategy to encourage home treatment for those with less serious

symptoms to reduce overcrowding in hospitals and also to contain the spread of pan-

demic flu by reducing the number of contacts between infected and noninfected indi-

viduals. A home treatment strategy would require a system to provide training and

support for home caregivers.

Some public health experts have argued that quarantine measures to combat pan-

demic flu will fail. As noted earlier, such measures worked well in slowing the spread

of SARS, but SARS is much less contagious than the flu and has a longer incubation

period. Unlike people with SARS, people with the flu can transmit the virus begin-

ning one day before symptoms develop. Less stringent social distance measures—clos-

ing public places such as schools, shopping malls, and movie theaters—may be more

effective because those measures do not require that infectious people be identified. As

time goes by and the flu dissipates, however, people may begin to resume normal ac-

tivities, which could lead to a second wave of flu infections.

44. Lister, Pandemic Influenza, p. 10.

Box 2.

“Surge” Medical Capacity of the Departments of

Defense, Veterans Affairs, and Homeland

Security

One of the resources that might be drawn on in the event of an outbreak of

pandemic influenza is the medical facilities and staff of the Departments of De-

fense (DoD) and Veterans Affairs (VA). Many of those individuals also partici-

pate in the Department of Homeland Security’s National Disaster Medical Sys-

tem (NDMS), whose teams of medical personnel support other professionals

in the event of a major emergency or disaster. In addition, the military has mo-

bile facilities that might provide hospital and medical capacity during a pan-

demic.

Department of Defense Facilities

The Department of Defense has more than 50 military hospitals and medical

centers in the continental United States, providing more than 3,500 staffed

beds for active-duty personnel and their dependents and for military retirees

and their families. On any given day, about 70 percent of those beds are occu-

pied, although many are used by patients seeking elective or nonurgent care

that could be deferred if beds were needed for flu victims. In addition, DoD

operates more than 400 outpatient clinics. In the event of a pandemic, those

facilities might be able to treat ambulatory cases, thus providing some addi-

tional capacity for treating flu victims. Staffing could be supplemented by call-

ing up medical personnel in the National Guard and Reserves. However, that

approach, rather than adding to the total number of medical personnel avail-

able nationally to treat avian flu cases, would instead merely shift medical per-

sonnel from the civilian to the military sector.

Facilities Operated by the Department of Veterans

Affairs

The Department of Veterans Affairs’ 157 hospitals provide over 18,000 acute

care beds. VA also operates more than 860 outpatient clinics. Those facilities

serve 7.5 million veterans who are enrolled in the VA medical system. Al-

though most of those facilities currently operate at high occupancy or utiliza-

tion rates, in the event of a pandemic outbreak, treatment of many patients

who were seeking elective or nonurgent care could be deferred if VA facilities

and personnel were needed. In addition, VA could increase its inpatient capac-

ity by reassigning personnel and adding beds where possible. However, VA is

likely to be faced with an influx of veterans seeking treatment for avian flu,

who may quickly fill VA hospitals to capacity.

Box 2.

Continued

The National Disaster Medical System

The National Disaster Medical System, a section within the Department of

Homeland Security, supports other federal agencies in responding to the medi-

cal aspects of major emergencies and federally declared disasters, including

natural disasters, major accidents, and events that involve weapons of mass de-

struction. The NDMS includes a Disaster Medical Assistance Team, National

Nurse Response Team, National Pharmacy Response Team, and other key sup-

port teams. Those groups treat victims in and responders to domestic disasters

by providing medical care at a disaster site, in transit, and in hospitals and

other treatment facilities that participate in the system.

Because many team members work in the medical field on a daily basis, the

teams do not necessarily add to the number of medical professionals who

would be available to treat avian flu cases. They would, however, add to the

number of people trained to work together to provide care and manage re-

sources in case of a disaster or emergency.

Mobile Medical Facilities

The military also has mobile medical facilities (which include the Army’s field

hospitals, the Navy’s hospital ships and expeditionary medical facilities, and the

Air Force’s theater hospitals) that are intended to provide medical and surgical

care to military forces deployed in combat zones. Those facilities are designed

to be set up rapidly and to be self-supporting. Many are prepositioned overseas.

The facilities are intended to be staffed by military medical personnel drawn

from fixed facilities (such as military hospitals) as well as by medical personnel

activated from the National Guard and Reserves. Although such resources

might be needed to care for military personnel, nondeployed mobile medical

facilities in the United States might be available to provide care to civilians in

case of an avian flu pandemic. One advantage of the mobile hospitals is that

they are self-sustaining, providing their own power and traveling with prepack-

aged sets of drugs, medical supplies, and equipment. Field hospitals might be

particularly useful if medical facilities were needed in remote areas without ac-

cess to power, water, and other services. Several thousand beds could be made

available by setting up these facilities in the United States, but the actual num-

ber would vary depending on deployment and training

schedules.

Options for the Longer Term

If an avian flu pandemic did not start for several years, the United States would have

many more opportunities to improve its response capability. Although monitoring

and source control could continue to be effective in helping to prevent a potential

pandemic, the possibility of new vaccine techniques, coupled with the effective use of

larger stockpiles of vaccines and antiviral drugs, might offer a way to ameliorate pan-

demics in the future. Within the next year or two, currently available drugs could be

produced in larger quantities and stockpiled. The major risk associated with that ap-

proach, however, is that those stockpiled agents will prove to be ineffective. Also of

concern is protection of the patent rights of existing producers of antiviral drugs to

ensure that incentives to develop new such drugs remain strong.

For the future, new drugs and methods of production, particularly for vaccines, could

be developed and put in place. However, a better long-term outlook may depend on

policies that are implemented in the very near term. Yet that approach has pitfalls as

well—for example, those associated with active government intervention in private

investment and production. Direct investment by the government in production fa-

cilities runs the risk of the government’s essentially picking winners and losers—which

it does poorly. Less risky alternatives might be to contract for research into new drugs

and production technologies, guarantee purchases, build the market for annual influ-

enza vaccinations, and address a number of issues that some observers contend are ob-

stacles to private investment, including liability protection for vaccine producers, and

patent protection for innovations. The measures listed above are not mutually exclu-

sive and could be put in place in conjunction with direct government financing of

production facilities.

Vaccines. Over the long term, policies could be pursued to help ensure a sufficient

supply of vaccine—in part by taking advantage of new technologies—and to develop

a more effective process for distributing it to the public. Over the next year, stockpiles

of vaccine against the H5N1 strain could be built, and the H5N1 strain could be

added to the regular annual flu vaccine. Later, new dose-sparing approaches and im-

munization techniques could be developed to increase the number of people who

could be inoculated with a fixed supply of vaccine raw material.

Build a Stockpile of H5N1 Vaccine. Under this policy, manufacturers would produce

H5N1 flu vaccine in the “off-season” (that is, the time of the year when they are not

producing at full capacity), using their existing technology and their capacity for pro-

ducing the seasonal flu vaccine. Nevertheless, vaccines stockpiled to inoculate people

against the H5N1 strain could turn out to be ineffective against a mutated strain, and

shortages would persist.

Some people have raised concerns that manufacturers will decline to participate in

government stockpile programs because Securities and Exchange Commission (SEC)

guidance requires revenue recognition to be delayed until the buyer takes delivery or,

in the case of vaccines, until they are delivered from the stockpile. However, on

December 5, 2005, the SEC released new guidance that allows companies to recog-

nize revenue for vaccine sales to federal government stockpile programs.

Vaccinate People with the H5N1 Vaccine. Vaccinating people with an H5N1 strain now

might provide some of the population with additional immunity in the event of an

H5N1 pandemic. However, a pandemic flu strain might be significantly different

from the current H5N1 strain, which would make advance vaccinations with the cur-

rent strain ineffective. Adding H5N1 to the regular seasonal flu vaccine would require

manufacturers either to make fewer doses of the seasonal vaccine or to remove one of

the current strains that it includes. Such a strategy might increase the likelihood that

more people would get the flu from the removed strain.

Develop Dose-Sparing Vaccines and New Immunization Techniques. New technologies

may be available in the near future that will have the potential to reduce the amount

of antigen needed for a vaccine to be effective. Promising dose-sparing technologies

include intradermal administration (injected between the layers of the skin) as op-

posed to intramuscular administration (injected within the muscle) and adjuvanted

vaccines. Adjuvants are substances that are included in the vaccine to make it effective

at a lower dose; they are routinely used in childhood vaccines. Currently, the United

States has no approved influenza vaccines that contain adjuvants. In October, Chiron

announced promising results from a clinical study that tested its investigational adju-

vanted vaccine against another avian influenza virus, H9N2. In that study, adjuvanted

vaccine doses containing as little as 3.75 μg of antigen proved highly capable of pro-

ducing an immune response.

Data from clinical trials are required to gain regulatory approval from the FDA for

those technologies. NIH has in place a network of vaccine and treatment evaluation

units (VTEUs) to quickly conduct clinical trials. Using its VTEUs, NIH completed

trials of a seasonal influenza vaccine manufactured by GlaxoSmithKline to accelerate

the FDA approval process and increase the supply of vaccine in the United States in

time for the 2004-2005 flu season.

NIH has also just completed trials comparing

intradermal with intramuscular administration using a candidate H5N1 vaccine man-

ufactured by sanofi pasteur and expects to have published results within the year. In

addition, trials of an H5N1 adjuvanted vaccine will begin soon.

45. That guidance is available at http://www.sec.gov/rules/interp.shtml.

46. The seasonal flu vaccine is a trivalent vaccine—that is, it contains antigens from three influenza

virus strains (two influenza A virus strains and one influenza B virus strain).

47. “Chiron Announces Promising Data from Clinical Study of Adjuvanted Avian Influenza Vaccine;

Results Confirm Previous Clinical Studies: Chiron’s MF59 Adjuvant Significantly Enhances

Immune Response,” Chiron News, October 28, 2005 available at http://phx.corporate-ir.net/phoe-

nix.zhtml?c=105850&p=irol-newsArticle&ID=774795&highlight=.

48. Personal communication to the Congressional Budget Office from NIH staff.

Strengthen Demand-Side Incentives. A government commitment to purchase influenza

vaccines that are effective against potential pandemic strains would be a strong market

signal to producers that investment in new capacity and technologies will be re-

warded. An increase in the demand for annual flu vaccine would amplify that signal.

Increasing the market for the seasonal flu vaccine would also provide the immediate

benefit of reducing the cost of the seasonal flu.

An example of demand-side incentives in the bioterrorism area for development of

vaccines and treatments against biological agents is Project BioShield, begun in 2004.

The program is designed to encourage such development through the use of advance

purchase agreements that offer the promise of a government sale to encourage drug

manufacturers to develop products that have limited markets with no commercial

application. (Without advance purchase, manufacturers might be reluctant to invest

in expensive clinical trials and production capacity.) To date, funds appropriated for

such purchases total $5.6 billion; about $3.4 billion of that total is currently available

for obligation. The government has entered into only a few contracts, the largest of

which ($800 million) is for an anthrax vaccine scheduled for delivery in 2007. Prod-

uct development has been slow, however. The government has paid out only $70 mil-

lion for the delivery of final products. Critics of this kind of approach question

whether the government should be providing funding to companies during this stage

of the development process and whether the approach provides a strong incentive to

manufacturers.

In terms of the market for the seasonal flu vaccine, the CDC currently recommends

that about 190 million people get a flu shot each year, and in recent years when sup-

plies were adequate, about 45 percent of that number were actually vaccinated. The

annual market might be expanded by intensifying efforts to get more individuals in

the recommended groups inoculated (for example, only about 40 percent of health

care workers were inoculated in 2003) or by including more groups among those for

whom vaccination is recommended—all college students, for instance. Yet the gains

from such a strategy might be limited: the government and the news media actively

promote flu shots now, and more of the same might not produce large increases in the

size of the market.

Another demand-expanding strategy for the seasonal flu vaccine would be for the

CDC to increase the recommended dosage particularly for the elderly, a group that is

a disproportionately large consumer of flu vaccine and one for which the standard

dosage of the seasonal flu vaccine generally provides a lower immune response. More

dramatically, the government could implement the recommendations of an Institute

of Medicine panel that in 2004 called for a federal mandate on insurers, accompanied

by offsetting subsides, to cover the cost of vaccines (influenza and others) under their

policies, as well as a vaccine voucher program for people who do not have health in-

surance.

49. Institute of Medicine, Financing Vaccines in the 21st Century: Assuring Access and Availability

(Washington, D.C.: National Academy Press, 2004), pp. 11-14.

Strengthen Supply-Side Incentives to Increase Capacity and Improve Technologies. The

Administration’s plan and the various Congressional proposals span a wide range of

supply-side stimulants that include additional funding and other incentives for re-

search and development and clinical trials, government support for construction and

renovation of facilities, streamlined regulatory procedures, and liability protection for

manufacturers.

B Provide financial assistance to enhance vaccine production capacity. An important part

of the Administration’s proposal to increase the nation’s capacity to produce influ-

enza vaccine is, as HHS Secretary Michael Leavitt testified, “financing the estab-

lishment of new cell-based vaccine manufacturing facilities.”

President Bush, in

his speech announcing the proposal, noted that the HHS plan included a request

for 2.8 billion dollars for cell-culture technology. Presumably, financing for pro-

duction facilities is included in that total. Also provided in the Administration’s

plan is financing for the retrofitting of egg-based manufacturing facilities.

Detailed information about financing, however, is not provided in the HHS plan;

thus, financing could involve a number of different mechanisms ranging from out-

right grants to loans (subsidized and unsubsidized) to loan guarantees. Govern-

ment financing might not take into account the value of production capacity as an

investment, and such choices by the government discourage outsiders or active par-

ticipants in the industry that wish to pursue a different approach. Consequently,

the use of demand-side incentives—such as stockpiling, guaranteed purchases, and

the like—gives market forces wider latitude.

B Establish incentives for research and development. Assistance from the government

for research and clinical trials might encourage manufacturers to apply for licenses

for new technologies if the approaches proved cost-effective. Both the Administra-

tion’s proposal and those of Members of Congress include such measures. New

technologies might comprise cell culture-based influenza vaccines that could cir-

cumvent possible problems presented by H5N1 strains that are lethal to chicken

embryos. That technology might be available in three to five years. Other ap-

proaches that are not as far along in their development include recombinant DNA

production and the development of a universal vaccine that would work against all

strains of the flu. Some scientists argue that a live-attenuated flu vaccine (that is,

containing live but weakened influenza viruses) or a whole-virus vaccine could pro-

vide a stronger immune response at lower doses. (MedImmune’s Flumist is a live-

attenuated flu vaccine for seasonal flu that is administered as a nasal spray.) Other

researchers argue that it would be too risky to create live-attenuated or whole-virus

pandemic flu vaccines.

Manufacturers may be unwilling to incur the regulatory costs of getting new tech-

nologies approved when prices for seasonal flu vaccines are low. NIH could, as it

50. Statement of Michael Leavitt, November 2, 2005, p. 6.

did in response to last year’s seasonal influenza vaccine shortage, use its network of

vaccine and treatment evaluation units to quickly conduct clinical trials of vaccines

made from new technologies and thus accelerate the FDA approval process, reduc-

ing regulatory costs for manufacturers. Policymakers could also provide tax incen-

tives, prizes, or other rewards to firms that developed certain technologies to

achieve stated goals.

B Provide liability protection. Faced with having their products used in a public health

emergency, manufacturers contend that the federal government should offer them

liability protection. Liability exposure associated with vaccine injuries is often cited

as one of the reasons for the small number of vaccine manufacturers and the lack of

innovation in the industry. As the need for protection from infectious diseases and

bioterror agents has grown, so have demands for broad federal shielding of manu-

facturers from the financial risk associated with vaccine production, particularly for

vaccines distributed to large numbers of healthy people in a public health emer-

gency.

However, there are questions about the extent to which potential liability exposure

discourages future investment in vaccines and vaccine technologies. Some observ-

ers view liability as a minor problem for vaccine manufacturers and assert that low

profit margins for certain vaccines, together with the lack of a dependable market,

are the main reasons companies do not invest in vaccines.

Those observers also

note that companies invest in other vaccine ventures (for example, vaccines to pro-

tect against HIV and cervical cancer) without extra liability protection. Opponents

of broad federal liability protection argue as well that individuals should have ac-

cess to judicial remedies in vaccine injury cases and that without an avenue to ad-

dress injuries, people might be deterred from seeking immunization.

The federal government has several options for addressing liability concerns of

manufacturers; they include limiting the circumstances in which individuals can

sue, indemnifying manufacturers, providing assistance with liability insurance, and

establishing or broadening vaccine injury compensation programs. An existing no-

fault compensation system for injuries related to childhood vaccines⎯the National

Vaccine Injury Program (VICP)⎯compensates individuals who are harmed by cer-

tain vaccines and protects manufacturers by limiting suits that can be pursued out-

side of the program. That program covers the seasonal influenza vaccine but not

vaccines against a flu pandemic. An additional program provides compensation to

health care workers who are injured by the smallpox vaccine. One option for ad-

dressing liability concerns about vaccines against pandemic influenza might be to

include them in the VICP. However, because the fund’s financing comes from ex-

51. American Trial Lawyers Association, “Liability Concerns Are Not Affecting Vaccine Pro-

duction,” ATLA Factsheets (November 3, 2005), available at www.alta.org/FACTS/products/

vaccineshortage.aspx.

52. See Congressional Budget Office, The Economics of U.S. Tort Liability: A Primer (October 2003),

for a broad discussion of liability issues.

cise taxes levied on covered vaccines, that avenue may not be the most appropriate

for a mass campaign against pandemic influenza. Instead, a separate compensation

fund similar to the smallpox fund could be set up to work alongside other strategies

to limit liability exposure for firms.

Antiviral and Other Drugs. A reason for the lack of availability of antiviral drugs is

weak current sales, in part a consequence of the very short window in time in which a

doctor can prescribe the drugs and expect them to be effective. By the time a doctor is

consulted by an infected patient with the seasonal flu, it is often too late for antiviral

drugs, hence the preference for flu vaccination. As with vaccines, the supply of cur-

rently available antiviral drugs will be greater, and new drugs will be more likely to be-

come available over time. For the next year or so, significant obstacles stand in the way

of increasing production of currently available drugs by enough to meet worldwide

demand (see Box 3).

Strengthen Demand-Side Incentives to Increase the Production of Currently Available An-

tiviral Drugs. Steady demand from the federal government to build a stockpile of anti-

viral drugs and influenza drugs purchased at a price that producers considered ade-

quate could provide greater incentives for companies to invest or deploy existing

resources to large-scale production of those agents. However, for manufacturers to in-

vest in and convert production to pandemic influenza drugs would probably require

assurances from the government regarding how much it will pay and for how long.

During an outbreak, the federal government might threaten to confiscate intellectual

property rights as a way of motivating manufacturers. During the anthrax attacks of

October 2001, for example, the Secretary of Health and Human Services extracted

deep cuts from Bayer for its antibiotic Cipro (paying only one-quarter of the market

price for the crucial drug) by threatening the company’s patent protection.

Repeat-

ing actions of that type, however, will make companies less likely to invest in research

and development for drugs against pandemics.

Over the long term, individuals and private entities making their own assessment of

the risk of pandemic flu may also contribute to higher demand and establish their

own stockpiles of antiviral drugs. Some observers argue that in the event of a pan-

demic, government distribution of Tamiflu and Relenza would be too slow so it

would be better to distribute them prior to a pandemic. Others caution that incorrect

dosing could be ineffective and that flu viruses with increased resistance to Tamiflu

could develop if people take the drug incorrectly.

Strengthen Supply-Side Incentives to Develop New Antiviral Drugs. The Administration’s

proposal includes support for research and development for new antiviral drugs. But

maintaining the intellectual property rights held by the current producers of such

53. “Fearing Avian Flu, Bioterror, U.S. Scrambles to Fill Drug Gap,” Wall Street Journal, November 9,

2005.

drugs is also likely to be as important as subsidies for new drug development in deter-

mining whether such activities actually occur.

On the research front, some researchers suggest that the government provide more

money for research and development and for clinical trials to support the develop-

ment of a new generation of antiviral drugs. Some observers have pointed out the ef-

fectiveness of statins (cholesterol-lowering drugs) in reducing flu symptoms from the

H5N1 virus. Yet even without additional government resources for antiviral research,

private firms are likely to view new antiviral drugs and antibiotics (some of which may

treat pandemic influenza as well as other diseases) as potentially profitable enterprises.

Improve Readiness of the Health Care System. Over the long term, the preparedness of

health care facilities to deal with a pandemic poses different challenges than those as-

sociated with purchasing vaccines and antiviral drugs. The construction of new hospi-

Box 3.

Obstacles to Increasing the Production of Cur-

rently Available Antiviral Drugs

As with influenza vaccines, manufacturing mass quantities of certain products

to address a flu outbreak could require substantial investment in facilities and

might strain existing production processes. According to Roche, the maker of

Tamiflu (also known as oseltamivir), production of that drug takes six to eight

months and requires scarce ingredients and specialized facilities.

The active in-

gredient in Tamiflu—shikimic acid—comes from the star anise, a rare Chinese

cooking herb. Researchers at Michigan State University have developed a syn-

thetic version of shikimic acid; however, the production process for that com-

pound requires using specialized fermentation equipment that companies may

want to use for producing other products instead.

The manufacturing process for oseltamivir includes other complicating factors,

such as the potentially dangerous use of sodium azide, a highly reactive chemi-

cal (also used to make automobile air bags inflate). Many companies, including

Roche, contract out that step in the production process.

Even if those production issues were resolved, questions would arise about

whether to depend on existing producers in the near term or to require com-

pulsory licensing so that generic manufacturers who claim they are willing

1. Andrew Pollack, “Is Bird Flu Drug Really So Vexing? Debating the Difficulty of Tamiflu,”

New York Times, November 5, 2005.

tal wings or clinics to add beds to the system would be a costly policy option. A way

to begin to address the issue would be to adopt measures that allowed facilities to

maintain the flexibility to use beds in a local stadium or community center, should an

outbreak occur. But those solutions are by their nature local ones and not easily ad-

dressed by the federal government. At the federal level, assistance could be provided

by continued stockpiling of medical equipment (such as ventilators and antibiotics)

and technical assistance to communities, clinics, and hospitals. Federal support might

also be used to increase the supply of health care personnel in short supply.

Box 3.

Continued

and technically able to produce oseltamivir could do so. Having multiple com-

panies produce antiviral drugs could increase worldwide supply and prevent

manufacturing problems from contributing to shortages in an emergency.

However, the global market in which drug production takes place complicates

intellectual-property issues. The Indian drug company Cipla announced in

October that it would begin manufacturing small amounts of generic Tamiflu

without a sublicense from Roche; Taiwanese researchers have already produced

small quantities of Tamiflu; and a Japanese company, Sankyo, has developed a

new version of Relenza (zanamivir, another drug for treating influenza virus).

Intellectual-property concerns weigh into companies’ decisions about whether

to invest in products; a weakening of intellectual-property laws could deter

firms from producing necessary antiviral drugs now and (equally important) in

the future.

Roche, however, is in negotiations with some generic manufactur-

ers to sublicense Tamiflu production.

2. For a discussion of the relationship between future innovation and the protection of intel-

lectual-property rights, see Congressional Budget Office, Copyright Issues in the Digital

Media (August 2004), pp. 21-26.

3. James Kanter, “Roche Offers to Negotiate on Flu Drug,” New York Times, October 19,

2005.

Technical Appendix

Calculating the economic effects of a potential influenza pandemic with any degree of

precision is extremely difficult. There do not appear to be any empirical estimates of

the effects of the three flu pandemics that occurred during the twentieth century.

However, it is possible to develop a rough estimate of the likely effects by examining

the possible effects on the supply side of the economy and then adding the effects of

declines in demand that result from people trying to avoid social contact in stores and

other public places.

Supply-Side Effect

To calculate the supply-side impact of a pandemic, CBO calculated the effect of the

loss of employee work days on GDP in the five sectors of the economy—nonfarm

business, farm, households, nonprofit institutions, and general government—using

average productivity per employee calculated in 2004 (the last full year for which data

are available). That calculation required assumptions about (a) the gross attack rate in

each sector, (b) the case fatality rate for each sector, and (c) the number of weeks that

the “average” infected worker would miss because they were sick. Combining those as-

sumptions with the level of employment in each sector, allowed CBO to calculate the

“lost employment” for the year in which the pandemic occurs.

That lost employment

is then multiplied by average productivity in the sector to compute the impact on

GDP.

Specifically, the number of infected workers in a given sector is calculated by multi-

plying the gross attack rate by the level of employment in 2004. The infected workers

are assumed to miss the “average” number of weeks of work; those who are assumed to

die from their illness—computed by applying the case fatality rate to the number who

take ill—are assumed to miss a full year of work. Clearly, those workers will not rejoin

the labor force, but the calculation assumes that the reduction in the labor force will

raise real wages, thus encouraging some people who were not part of the labor force to

join it and eventually allowing the level of employment to regain its previous trend.

The specific values for the assumptions are given in Table A-1. For the severe scenario,

CBO assumed that, on average, 30 percent of the workers in each sector (except for

the farm sector) would take ill and, of those workers, 2.5 percent would die. CBO as-

sumed that those who survived would miss 3 weeks of work either because they were

sick or because they needed to care for family or friends that became sick.

Given the

less social nature of work in the farm sector, CBO assumed a milder impact: one-

tenth of the workers are affected and survivors miss only a single week of work (the

1. Implicit in the exercise is the assumption that the pandemic runs its course within 12 months, a

span that could occur in more than one calendar year.

2. Three weeks out of work is likely to be at the high end of the range for those who are infected.

However, it is also meant to account for healthy workers who are absent because they are caring for

sick family members or children who are home from school and because they are too fearful of

becoming sick to leave their homes.

Table A-1.

Assumptions Underlying Estimates of the Supply-Side

Impact of an Avian Flu Pandemic

Source: Congressional Budget Office.

Note: The gross attack rate is the percentage of the population that is infected with a disease. The

case fatality rate is the percentage of infected persons who eventually die from the disease or

complications.

case fatality rate was unchanged). Using data for 2004, CBO used average productiv-

ity per worker, by sector, to compute the impact on GDP of the employment lost to

the pandemic.

For the mild pandemic CBO assumed a 25 percent attack rate (except in the farm sec-

tor, which was assumed to be 5 percent), a case fatality rate of just over 0.1 percent,

and cut the time out of work to one-quarter of the duration assumed for the severe

scenario (i.e., just under four days absent, on average).

Demand-Side Effect

To calculate the demand-side effect, CBO examined GDP by industry and assumed

different declines in demand for different industries, based on judgments about the

degree of social interaction required in different industries. Given that there is little

historical evidence available to form these estimates, they are admittedly extremely

rough. Industries that require interpersonal contact are assumed to have the largest de-

clines in demand. For example, CBO assumed that demand would fall off by 80 per-

cent (for three months) in the entertainment, arts, recreation, lodging, and restaurant

industries, a set of industries that composed just under 4 percent of GDP in 2004 (see

Table A-2). Other industries were assumed to suffer a smaller decline in demand. Re-

tail trade, for example, was assumed to suffer a 10 percent decline, as were the whole-

sale trade and manufacturing industries.

In contrast, there would be a surge in de-

mand for medical care, which CBO assumed would rise 15 percent relative to a base

case without a pandemic. For the mild scenario, CBO assumed that the demand-side

declines in each industry were one-quarter of the declines in the severe scenario.

3. The analysis ignores the possibility that productivity among workers who remain on the job would

be likely to rise.

4. The government sector would not have any demand-side impact in addition to the supply-side

impact.

Economic Sector Severe Mild Severe Mild Severe Mild

Nonfarm Business 30 25 3 0.75 2.5 1.14

Farm 10 5 1 0.25 2.5 1.14

Household 30 25 3 0.75 2.5 1.14

Nonprofit Institutions 30 25 3 0.75 2.5 1.14

Government 30 25 3 0.75 2.5 1.14

(Percent) Weeks Out of Work (Percent)

Case Fatality Rate Gross Attack Rate

Note that some of the impact on overall GDP of the drop in demand would already

be accounted for by the decline in supply. Therefore, CBO subtracted the supply-side

impact of the pandemic on each industry from the estimate of demand-side impact

before adding the two effects together to calculate the effect on GDP. That procedure

avoided double-counting the supply-side effects.

Table A-2.

Assumed Declines in Demand, by Industry, in the

Event of an Avian Flu Pandemic

(Percent)

Source: Congressional Budget Office.

Note: The severe scenario describes a pandemic that is similar to the 1918-1919 Spanish flu out-

break. It incorporates the assumption that a particularly virulent strain of influenza infects

roughly 90 million people in the United States and kills more than 2 million of them. The mild

scenario describes a pandemic that resembles the outbreaks of 1957 to 1958 and 1968 to

1969. It incorporates the assumption that 75 million people become infected and about

100,000 of them die from the illness or complications.

Private Industries

Agriculture 10 3

Mining 10 3

Utilities 0 0

Construction 10 3

Manufacturing 10 3

Wholesale trade 10 3

Retail trade 10 3

Transportation and

67 17

Information (Published,

Finance 0 0

Professional and

Education/health care

Education 0 0

Health care -15 -4

80 20

governmen

Government

broadcast)

business services

accommodation/food

warehousing

Air

Rail

Mild Scenario

Other services except

Federal

State and local

Severe Scenario

Arts and recreation

Accommodation

Food service

Arts/entertainment/

Transit