The objective of this modeling exercise was to evaluate the potential effectiveness of epidemic control strategies that might be deployed in response to a bioterrorist attack. Our main finding is that contact tracing and vaccination of household, workplace, and school contacts, along with effective isolation of diagnosed cases, can control epidemics of smallpox. In our 6,000-person town model, we found that in scenario 3 (the combination of interventions most closely parallel to current U.S. governmental policies) the expected total number of smallpox cases that would ensue from ten simultaneous introductions would be 25–40 additional cases (CDC, 2002). Our findings in the 50,000-person town model were consistent with these estimates; under scenario 3,500 introductions into a population of 50,000 would give rise to approximately 1,100 new cases of smallpox. In both size versions of the model, reactive mass vaccination at the town level had additional value in bringing an epidemic under control. We estimate the number of reactive mass vaccinations required to incrementally reduce the epidemic by one case to be about 190 vaccinations in the 6,000-person town/10-attack-case model versions and about 35 vaccinations in the more intense 50,000-person town/500-attack-case model version.
Although a good deal of variation in the size and other characteristics of the modeled epidemics is expected in a highly stochastic epidemic model, we were nonetheless surprised by some of our observations (Bailey, 1953; Whittle, 1955). In our epidemic simulation runs, 1) epidemics ranged dramatically in size and duration based on chance alone, 2) the epidemic impact of individual index (attack) cases ranged from no transmissions whatsoever to large and lengthy transmission chains, and 3) the epidemic reproductive rate varied substantially by clinical disease type and by epidemic generation and was dependent on the underlying social network configuration. These results suggest that the heterogeneity of our microscale, agent-based model has significantly impacted the resultant epidemics.
It is possible that some important parameters may not have been considered in the development of this model. For example, age-specific differences in the pathogenicity and transmissibility of smallpox were not considered, other than as they relate to age older than 32 years and prior vaccination status as well as social contact processes (schools for children vs. workplace for adults). We did not explicitly include risks of smallpox vaccination as a source of adverse outcome in our model. The number of vaccinations used in each modeled response is given in the Results section and can be used to estimate adverse outcomes. Another potentially important biological variable unexamined in this exercise is the effect of seasonality on transmission of smallpox (Fenner et al., 1988).