From the 1Yale School of Management, and Department of Epidemiology and Public Health, Yale School of Medicine, New Haven, Connecticut, and the
2Graduate School of Business, Stanford University, Stanford, California.
Address correspondence to: Edward H. Kaplan, Yale School of Management, Box 208200, New Haven, Connecticut 06520–8200; firstname.lastname@example.org
Submitted 4 September 2002; final version accepted 14 October 2002.
Editors’ note: An invited commentary on this article appears on page 93.
The global eradication of smallpox stands among the most important public health achievements of the last century. 1 However, fears that smallpox bioweapons 2,3,4 have fallen into the hands of rogue states or terrorist organizations have rekindled interest in the methods used to control smallpox outbreaks. Surveillance-containment employing ring vaccination (whereby symptomatic smallpox cases were sought and isolated and their surrounding contacts traced and vaccinated) has been widely acknowledged as the key to smallpox eradication, 1 though some have questioned this claim. 5 In the United States, where the interim policy in the event of a smallpox attack is based on ring vaccination, 6 we have argued that localized mass vaccination from the moment an attack is recognized would result in far fewer casualties than ring vaccination, including deaths that result from vaccination itself. 7
In this commentary, we closely examine a figure based on historical data reported by Foege et al. 8,9 documenting smallpox control and eradication in West and Central Africa. This figure has been resurrected as empiric proof of the effectiveness of ring vaccination, and has been presented in important meetings addressing smallpox response policy in the United States. 10 It is important that data offered to garner support for a policy position stand on their own as presented, and indeed it was the public presentation of this figure that prompted our investigation. Contrary to the claim made, we will argue that the reported decline in smallpox cases matches what one would have expected on the basis of increased vaccination coverage alone. Herd immunity, not surveillance-containment, appears to be the real story behind the eradication of smallpox in West and Central Africa.
The key result of Foege et al. presented in support of ring vaccination was reproduced from Figure 9 of their paper in the Bulletin of the World Health Organization8 (the same figure first appeared as Figure 2 in reference 9). We have reproduced this figure as shown publicly as Figure 1. 10,11 This figure reports the ratio of reported to “expected” cases of smallpox both before and after the initiation of surveillance-containment in West and Central Africa, as well as the percentage of the population that was not vaccinated. Though these data were aggregated over age and over 20 different countries, the intent of the figure is clear: note the seemingly sudden decline in the ratio of reported to expected cases after the surveillance-containment activities began. Indeed, with reference to this figure, Foege et al. report that “the results were dramatic.”8,p.218
However, closer examination raises several questions about the observed decline. First, note the logarithmic scale on the vertical axis. To the untrained eye, the drop from 100% to 10% appears equivalent to the drop from 10% to 1%. The clear impression is that although the percentage of the population that was not vaccinated declined slowly over time, the ratio of reported to expected smallpox cases fell like a rock shortly after surveillance-containment began. In fact, vaccination coverage tripled—from 20% to 60%—between January 1968 and March 1969. This change in vaccination coverage is visually eclipsed by the February through March 1969 drop in the reported-to-expected smallpox case ratio, yet the latter decline was only nine percentage points. Why was a logarithmic scale employed to report these data?
Second, expected smallpox cases were computed without accounting for the tripling in vaccination coverage just noted. Rather, the authors considered the average number of cases reported monthly from 1960–1967 (reported as Figure 3 in reference 8 and reproduced as Figure 2 here) as an indication of the number of smallpox cases that should be expected. Note in particular the increasing trend in monthly reported cases from January through April. That January through March are the last 3 months included in computing the ratio of reported to expected cases further exacerbates the illusion that smallpox cases declined much more than expected. There are generally two things that can go wrong when computing a ratio: the numerator and the denominator. In the present application, the denominator is simply a historical average of past cases, ignoring the important tripling in vaccination coverage that actually occurred.
Taken together, Figures 1 and 2 (Figures 9 and 3 in reference 8) enable one to determine the actual number of smallpox cases reported in West and Central Africa over the time period of interest. The actual reported cases along with the percentage of the population that was not vaccinated as deduced from Figure 1 appear in Figure 3. Here there is no mystery. More or less in lockstep over time, both the fraction of the population that was unvaccinated and the reported number of smallpox cases declined.
A different view of these same data is shown in Figure 4, where for each month we have plotted the reported number of smallpox cases against the fraction of the population that was vaccinated. The implication is very clear: as vaccination coverage tripled from 20% through 60%, monthly smallpox cases in West and Central Africa declined from approximately 800 to 30.
In contrast to Figure 1, it is no longer possible to identify September 1968 (or thereabouts) as the initiation date of surveillance-containment from Figures 3 and 4. Worded differently, once the increase in vaccination coverage is taken into account, there is little left to explain in the reported pattern of smallpox cases over time. The relation between vaccination coverage and the number of reported cases is the same before and after September 1968. One could argue that the surveillance-containment activities were required to obtain the tripling in vaccination coverage achieved, but the clear conclusion remains the same: it was the tripling of vaccination coverage that contributed most to the eradication of smallpox in this part of the world.
That the decline in smallpox cases can be explained plausibly by increased vaccination coverage as opposed to surveillance-containment is an important lesson to understand as smallpox control methods are revisited in the current bioterror policy debate. Before surveillance-containment activities began in September 1968, 45% of the population had been vaccinated, although survivorship from past smallpox outbreaks implied even higher levels of immunity. By contrast, in the United States and many other countries today, there is virtually no immunity to smallpox. Although we have argued previously that ring vaccination would work better in areas with higher rather than lower levels of population immunity, 7 the analysis reported herein casts serious doubts on the value added by surveillance-containment over mass vaccination even in areas where nearly half of the population had been vaccinated.
These results strengthen our conviction that in the event of a smallpox bioterror attack, rapid mass vaccination in the area of the attack with the goal of reaching herd immunity as quickly as possible should be preferred in favor of a ring vaccination strategy. Moreover, because bioterror-induced smallpox outbreaks would likely lead to cases worldwide, planning should begin now to ensure that rapid vaccination could be implemented immediately in inflicted regions around the globe. It is therefore important that policy makers in developed countries and at the World Health Organization formulate a strategy for coping with global outbreaks when devising their vaccine stockpiling and distribution policies.
1. Fenner F, Henderson DA, Arita I, Jezek Z, Ladnyi ID. Smallpox and Its Eradication. Geneva: World Health Organization, 1988.
2. Alibek K. Biohazard. New York: Random House, 1999.
3. Miller J, Engelberg S, Broad W. Germs. New York: Simon and Schuster, 2001.
4. Tucker JB. Scourge: The Once and Future Threat of Smallpox. New York: Atlantic Monthly Press, 2001.
5. Anderson RM, May RM. Infectious Diseases of Humans: Dynamics and Control
. Oxford: Oxford University Press, 1991;especially p.90.
7. Kaplan EH, Craft DL, Wein LM. Emergency response to a smallpox attack: the case for mass vaccination. Proc Natl Acad Sci USA 2002; 99: 10395–10440.
8. Foege WH, Millar JD, Henderson DA. Smallpox eradication in West and Central Africa. Bull World Health Organ 1975; 52: 209–222.
9. Foege WH, Millar JD, Lane JM. Selective epidemiologic control in smallpox eradication. Am J Epidemiol 1971; 94: 311–315.
10. Margolis HS. Smallpox vaccine performance: efficacy, effectiveness and vaccination strategies. Presented at: the Advisory Committee on Immunization Practices, Smallpox Work Group Meeting; 8 May 2002; Atlanta, Georgia.
11. Margolis HS. Smallpox control strategies and vaccine availability. Presented at the Institute of Medicine Forum on Smallpox: The Scientific Basis for Vaccination Policy Options, National Academy of Sciences; 15 June 2002; Washington DC.