The BW Protocol as a Health Care Intervention

The Biological Weapons Convention (BWC) of 1972 prohibits the development and stockpiling of biological weapons. Because the BWC has no means to verify compliance with the Convention, it has not prevented the proliferation of biological weapons even in some countries that are parties to the Convention. The international community recently had an unusual opportunity to strengthen the BWC by enacting a protocol, which an Ad Hoc Group of States Parties began negotiating in Geneva in 1996. The negotiations resulted in a compromise proposed protocol in 2001, which requires investigations and on-site visits within the geographic boundaries of the States that are party to the protocol.¹ While there is wide support among nations for a protocol, the United States has rejected the proposed protocol on grounds that it would be ineffective in preventing BW development and use and would compromise national security by revealing US defensive strategies for BW. Unfortunately, the US has blocked enactment of any protocol for the time being. The following summary of a longer analysis on the FAS website deals with effectiveness of the proposed protocol.

The analysis considers the protocol to be a health care intervention that helps to prevent sickness and death from a biological weapons attack. This perspective allows us to use standard pharmacoeconomic procedures and benchmarks to analyze the cost-effectiveness and cost-benefits of the proposed protocol. Pharmacoeconomic analyses are widely used to look at the costs, benefits and effectiveness of health-care interventions to assess health care choices, weighing costs and outcomes of new drug therapies and prevention practices. Each choice has its cost-and relative beneficial outcome. Health care analysts use pharmacoeconomics to determine the best "health-care buy." Seen in the light of quantitative pharmacoeconomics, our analysis shows that a protocol that would provide only a small decrease in the probability of a BW attack on the US would still be highly cost-effective, because its cost is so low compared to the extremely high cost of a successful BW attack of moderate size. Given the present terrorist threat, many believe that the risk of attack is not insubstantial. For a successful terrorist attack, it is more likely that weaponized BW agents will purposefully or inadvertently come from a State than be developed by terrorists on their own.

Data for the Analysis and Data Strategy²

To make a case for US support of the BWC protocol, assumptions regarding the data are conservative, that is, they tend to support the US government's contention that the protocol is ineffective. Thus, throughout the analysis, "conservative" means: where good data are not available or are uncertain, data values are chosen that tend to make supporting the protocol less favorable, financially and otherwise. The analysis is carried out for a period of 25 years.

Cost to the US to support the protocol. The total yearly cost to support the protocol was calculated to be $8.7 million. It is the sum of three costs: the 22% US contribution to the estimated $30 million yearly operating costs of the international organization to administer the protocol,³ $2 million yearly operating costs of the US implementing organization, and a $60,000 per year cost to facilities to host visits from the international organization.

Statistics for the scale of the attack. A BW attack on the US of moderate size is defined here as one in which 30,000 people are infected with 9,000 fatalities. This attack size is used throughout the analysis. A well-executed anthrax attack on a US city could cause over 100,000 fatalities,⁴ and a recent simulation of a small pox attack on the US called "Dark Winter" estimated that several million people could be killed.⁵ Thus, the BW attack size of this analysis is conservative compared to what we could experience.

	Seen in the light of quantitative pharmacoeconomics, our analysis shows that a protocol that would provide only a small decrease in the probability of a BW attack on the US would still be highly cost-effective.

For a small pox attack, 30 percent of the victims could die.⁶ For an anthrax attack, a higher percentage of victims could die.⁷ Since each death costs the US about $1.2 million (see the economic value of human lives, calculated below) for an attack with an agent that causes greater than 30% fatalities, the cost will rapidly rise with the number of fatalities. It is assumed that the average age of victims of a BW attack is 42 years, slightly over the average age in the US of about 37 years.⁸ Average age and life span are needed to determine life-years saved for cost-effectiveness analysis.

The economic value of human lives. The value of human lives and lives saved by a health-care intervention can be expressed in dollars. In the US, a human life may be valued at $34,000 per year, which is the gross national product per year (GNP) per capita. For an average victim's age of 42 years and with a life expectancy of 77 years,⁹ each life saved through prevention of a BW attack results in 77 _ 42 = 35 life-years saved. Thus, the average dollar value of each life saved from preventing a BW attack is 35 x $34,000 = $1.19 million. This number is used in the cost-benefit analysis.

Data for the likelihood of an attack and protocol effectiveness. The conservative probability of 0.01 per year for an attack on the US is chosen for the analysis. This low probability translates to a 50 percent chance of no BW attack of moderate size on the US in the next 69 years (years=log[0.5]/log[1-0.01]).

Effectiveness of the protocol is represented simply by the reduction of the probability for a BW attack.¹⁰ For a weakly effective protocol, we assume this reduction to be 10%; that is, the probability of a BW attack becomes 0.009 per year (0.01 x [1-0.1]). Said another way, a weakly effective protocol would prevent only one attack in ten. This is the value for reduction in attack probability used in the analysis.

All attack-associated data are weighted by the probability of attack without the protocol or with the protocol in place, as appropriate.

Direct and indirect costs of the attack. The direct cost of an attack is taken to be the cost of medical treatment of victims and is estimated to be $535 million, based on treatment costs for a number of serious diseases.¹¹

For this analysis, indirect costs of an attack are equal to direct costs for medical care of victims. This is a serious underestimate as the indirect costs will likely be several times the direct costs. Indirect costs would include prophylactic antibiotics or vaccines for those that are exposed, but not infected; lost work days of non-infected people who are ordered to stay at home; the cost of quarantining those who are infected or possibly infected; costs of environmental clean-up; costs related to mental or physical care of panicked unaffected individuals; criminal investigation costs; and costs of emergency and military responses to the attack. Thus, the total cost estimate ($535 million x 2) of about $1 billion, for the BW attack of moderate size is very conservative.

A third type of cost is the cost of fatalities. At an economic value of $1.19 million per person (see page 7), a BW attack with 9,000 fatalities would cost the US $10.7 billion, an amount that also would greatly exceed the direct costs.

Observations from the data. Compared to the small financial cost to the US to support the protocol, $8.7 million per year ($122 million net present value for 25 years), a moderately-sized attack on the US would be extremely expensive, minimally $1 billion, not including the $10.7 billion economic value of lost human lives. The logic of this extreme situation is almost akin to Pascal's logic in his well-known wager for belief in God: "Let us weigh the gain and the loss in wagering that God is… If you gain, you gain all; if you lose, you lose nothing. Wager, then, without hesitation that He is."¹² Like Pascal's argument, there is much to gain and little to lose in supporting the protocol. The simple pharmacoeconomic model for the analysis presented below "transforms" political arguments into a discussion of probabilities, which may help clarify discussion about effectiveness of the protocol.

A Primer on Pharmacoeconomics¹³

Two common pharmacoeconomic measures, cost-effectiveness and cost-benefit, were employed for the analysis. Cost-effectiveness (CE) is the dollar cost of the intervention divided by the life-years saved (lys) from the intervention:

Equation (1)

CE = COSTS/lys

For the BWC protocol, COSTS is net present value (NPV)¹⁴ cost of the protocol to the US over the years that the protocol is in place. lys is the life-years saved from the prevention of a biological weapons attack due to the protocol. What constitutes a cost-effective or good health-care buy? Health-care interventions with CE less than the GNP per capita, $34,000 per lys, are cost-effective and a good health-care buy. (In certain cases, society will tolerate higher cost per life-years saved for some health-care interventions.) In cost-benefit (CB) analysis, both BENEFITS and COSTS are defined and measured in dollar amounts.

Equation (2)

CB = BENEFITS-COSTS

BENEFITS are the cost-savings from prevention of a BW attack due to the protocol, which when in place, reduces the probability of an attack and thus reduces the probability-weighted direct and indirect costs of an attack and number of fatalities. All benefits are also expressed as NPV.

	Table 1. Data used in pharmacoeconomic analysis (Attack-related costs are realized only in the years when an attack on the US occurs).   No Protocol Protocol in Place Yearly Probabilities:     Attack 0.010 0.009 No Attack 0.990 0.991 Yearly Cost:     Protocol Implementation $0 $8.7 million Attack Related Costs:     Direct Cost of Attack $535 million $535 million Indirect Cost of Attack $535 million $535 million Cost of Fatalities $10,700 million $10,700 million

The model and results

In our simple model for pharmacoeconomic analysis, the probability of an attack in each year is independent of whether or not an attack occurred in the preceding year. Also, the costs of the protocol and of an attack are assumed to be constant over the 25 years for which the analysis is carried out. The cost and probability data for the analysis are summarized in Table 1.

The particular scenario represented by the probabilities in Table 1 is for a relatively ineffective protocol. To reiterate, effectiveness of the protocol is represented solely by reduction of the probability of an attack-in this case a 10% reduction from 0.01 to 0.009 per year for a BW attack with no protocol in place.

Summary of results of cost-effectiveness and cost-benefit analysis. The expected number of lives saved by the protocol is simply the difference in fatalities for no protocol vs. protocol in place. This difference is calculated as follows. The expected number of fatalities in any single year with no protocol is given by:

[Probability of an attack (no protocol)] x [number of fatalities] = 0.01 x 9000 = 90

The expected number of fatalities in any single year with the protocol in place is given by:

[Probability of an attack (Protocol in place)] x [number of fatalities] = 0.009 x 9000 = 81

Thus, in any one year, the expected number of saved lives from the protocol is 90 - 81 = 9 and the expected total number of life-years saved is therefore 9 x 35 = 315. The NPV for life-years saved over the 25 years of the analysis is 4,440 lys.¹⁵

The NPV cost of the protocol over the 25 years of the analysis is $122 million, so the cost-effectiveness of the protocol using Equation (1) is

CE = $122 million/4,440 lys = $27,500 per lys

which is under the $34,000 value demarcating a good health-care buy. More importantly, the value for CE was arrived at by using only assumptions about the data that were unfavorable to effectiveness of the protocol.

Using appropriately discounted and probability-weighted values in equation (2), CB = $43 million. Therefore, by supporting the protocol the US will actually save money; that is, the expected cost-savings are well above the cost of supporting the protocol for 25 years.

	By supporting the protocol the US will actually save money; that is, the expected cost-savings are well above the cost of supporting the protocol for 25 years.

A less conservative case

For a BW attack of greater scope or with higher indirect costs, the protocol could be considerably less effective, and still be a good heath-care preventative measure. The indirect costs of an attack could easily be ten times the direct costs, which may still be conservative. Furthermore, the protocol could be more effective and thwart 20% of all potential attacks. For this less conservative case, how low can the probability of attack be to ensure that the protocol is still cost effective and has positive benefits over costs; that is, CE is less than $34,000 cost per lys and CB is greater than $0?

For this case, the yearly probability of attack on the US can be as low as 0.004, which translates to a 50% chance of no BW attack on the US in the next 173 years. Thus, the protocol would be cost effective even if the risk of attack on the US is very small. Benefits over costs remain positive as well, CB = $65 million NPV.

Conclusion

Pharmacoenonomics is one analytical tool to gauge the effectiveness of the protocol. Based on this approach, it is in the US' interest to support a BWC protocol. The US' arguments for rejection of the proposed protocol-on the grounds of effectiveness-are without merit.

Dr. Lynn Klotz is a member of the FAS Working Group on Biological Weapons. He is a an expert in many areas of biotechnology and biological weapons control and has published several reports dealing with the Biological Weapons Convention Protocol, including papers on the pharmaceutical and biotechnology industries' response to the protocol and technical issues related to compliance with the BWC. He received the Dreyfus Teacher-Scholar Award for excellence in teaching, while an associate professor of biochemistry and molecular biology. The Gene Age: Genetic Engineering and the Next Industrial Revolution, which he co-authored with Edward J. Sylvester, was nominated for the Pulitzer Prize for Nonfiction in 1983.

NOTES:

1. See for example: Rosenberg, B., "Allergic Reaction: Washington's response to the BWC Protocol," Arms Control Today, July/August 2001. Link: http://www.armscontrol.org/act/2001_07-08/rosenbergjul_aug01.asp
2. The detailed description of the data and analysis can be found in the FAS Working Group Papers section of the FAS Chemical and Biological Arms Control Project web site (http://www.fas.org/bwc/papers/protocol_ce.htm) in the paper entitled "The Biological Weapons Protocol as a Health Care Intervention: Cost-effectiveness and Cost-benefit." The spreadsheet developed for this analysis allows testing of many scenarios and is available from Lynn Klotz lynnklotz@compuserve.com
3. Details of the operating budget and a spreadsheet for the budget may be found on the FAS web site. (http://www.fas.org/bwc/papers/structure/start.htm). The US 22% contribution is from the UN scale of assessment.
4. Christopher, G., et al., "Biological Warfare: a Historical Perspective," Journal of the American Medical Association (JAMA), August 1997. U.S. Congress, Office of Technology Assessment, Proliferation of Weapons of Mass Destruction: Assessing the Risk, OTA-ISC-559 (Washington, DC: U.S. Government Printing Office, August 1993). Link: http://www.wws.princeton.edu/~ota/disk1/1993/9341_n.html pg 54.
5. Disease by Design: De-mystifying the Biological Weapons Debate. Crowley, Mi. BASIC Research Report 2001.2, November 2001. Link: http://www.basicint.org/BWreport.htm Part II, Section 7.3.
6. Fatalities of 30% is commonly accepted for smallpox. See, for example, the links http://www.hopkins-biodefense.org/pages/agents/agentsmallpox.html and http://www.who.int/emc/diseases/smallpox/faqsmallpox.html
7. See, for example, http://www.fda.gov/cber/vaccine/anthrax.htm or Meselson, M., Guillemin, J., Hugh-Jones, M., et al. "The Sverdlovsk Anthrax Outbreak of 1979." Science 1994;266:1202-1208.
8. Reference: Mean Age of Americans: 36.6 (Nov. 1, 2000). Population Estimates. Program, Population Division, U.S. Census Bureau, Washington, D.C. 20233. Internet Release Date: January 2, 2001. Link: http://eire.census.gov/popest/archives/national/nation2/intfile2-1.txt
9. Reference: Life Expectancy of Americans: 76.7 yrs (1998). Anderson, R.N., National Vital Statistics Reports; Vol. 48, No. 18. Hyattsville, Maryland: National Center for Health Statistics. 2001. Link: http://www.cdc.gov/nchs/fastats/lifexpec.htm
10. Some opponents to the protocol would argue that the protocol could increase the probability of attack, not reduce it, because it would give a false sense of security or reveal our defensive strategies. Almost all protocol supporters understand that the protocol is only one of a number of measures needed to reduce the risk of biological weapons, so they would not have a false sense of security nor reduce efforts in other areas to prevent BW proliferation. Regarding revealing defensive strategies, the managed access procedures built into the proposed protocol are more than adequate to protect confidential national security information, as the US has stated to be the case in the Chemical Weapons Convention.
11. Hospital cost and stay estimates were derived from several different severe illnesses: 1) Respiratory syncytial virus (RSV) among the elderly costs $11,000 per RSV pneumonia hospitalization (Han, L.L., Alexander, J.P., Anderson, L.J., J Infect Dis 1999 Jan;179(1):2530); 2) Costs for treating viral meningitis are over $2,000 per day, but hospital stays are short, 4 to 5 days. (Parasuraman, T.V., Deverka, P.A., Toscani, M.R., Managed Care 2000 Jan;9(1):41-6); 3) For severe sepsis, the average costs per case were $22,100 (Angus, D.C., Linde-Zwirble, W.T., Lidicker, J., Clermont, G., Carcillo, J., Pinsky, M.R., Crit Care Med 2001 Jul;29(7):1303-10); 4) Estimate three weeks in hospital for small pox survivors (Forman, J., Boston Globe, Tuesday November 20, 2001, page C1).
12. From the internet version of the Stanford Encyclopedia of Philosophy (http://plato.stanford.edu/entries/pascal-wager).
13. See for example: Luce, B.R., and Elixhauser, A., "Socioeconomic Evaluation and the Health Care Industry," Journal of Research in Pharmaceutical Economics, Vol. 2(4) 1990.
14. Throughout the analysis, the interest or discount rate used to calculate NPV is 5%.
15. Whether we discount life-years saved is a philosophical and psychological question. One could argue that our children's and grandchildren's lives in the future are as valuable as our lives now, in which case we would not wish to discount life-years saved. In this analysis, life-years saved are discounted at 5%, the same as money. Since most of us believe that future lives are more valuable than future money, discounting future lives at the same rate as money is a conservative assumption, because it leads to higher cost/lys values making less of a case for the protocol.