Statistical Models and Bioterrorism: Application to the U.S. Anthrax Outbreak
Brookmeyer, Ron, Blades, Natalie, Journal of the American Statistical Association
The U.S. anthrax outbreak of 2001 resulted from letters that were intentionally contaminated with anthrax spores and sent through the postal system. Five deaths were associated with the outbreak. Over 10,000 persons nationwide were recommended to receive at least 60 days of antibiotics to prevent disease. The question arises as to how large the outbreak would have been if public health officials had not treated thousands of persons with antibiotics to prevent disease. At first it would appear that the question is unanswerable because of the paucity of data. Indeed, the number of cases of anthrax was small, and there is little epidemiological data from other human outbreaks because anthrax is rare in modern times. The objective of this article is to describe how statistical models and reasoning can help answer such questions about the 2001 U.S. anthrax outbreak and can improve the response to future acts of bioterrorism. Some of the substantive conclusions from these models have been previously described in Brookmeyer and Blades (2002). The present article focuses on the underlying statistical methodology and presents various extensions of the models, sensitivity results, and simulation studies.
Anthrax is caused by the bacterium Bacillus anthracis (Brachman 1980; Cieslak and Eitzen 2000). Inhalational anthrax, a serious and potentially lethal disease, occurs when an individual breathes anthrax spores through the lung (Inglesby et al. 1999). There is no person-to-person transmission of inhalational anthrax. Only 18 cases of inhalational anthrax were reported in the United States from 1900 to 1978.
On October 2, 2001, a 63-year-old male photo editor at a media publishing company in Florida was admitted to a hospital emergency department, complaining of nausea, vomiting, and fever (Jernigan et al. 2001). The symptoms began several days earlier on a recreational trip to North Carolina. An astute clinician quickly made the surprising diagnosis of inhalational anthrax. A second case occurred in a 73-year-old man who delivered mail at the same Florida company. The source of the anthrax spores was thought to be an intentionally contaminated letter because some employees reported seeing a suspicious letter on or about September 19, 2001 (Bush, Abrams, Beall, and Johnson 2001). That letter has never been recovered and the date of exposure to anthrax spores among these Florida cases has not been definitely ascertained. Following the confirmation of these two cases of inhalational anthrax, Florida health officials gave prophylaxis antimicrobials to approximately 1,100 persons who worked or visited the Florida media publishing company.
A letter addressed to Senator Tom Daschle and postmarked on October 9, 2001, was processed at the Hamilton post office in New Jersey and subsequently sent to the Brentwood post office in the District of Columbia on October 12. When the letter was opened in the Hart Senate Office Building on October 15, a white powder was released, which was subsequently determined to contain anthrax spores. The Hart Office Building was immediately closed, and all workers in the building were treated with antimicrobial prophylaxis. Initially, public health officials did not realize that postal workers who had handled sealed envelopes filled with anthrax spores were also at risk of disease. However, when cases of inhalational anthrax surfaced among postal workers at both the Hamilton and the Brentwood postal facilities, the postal facilities were immediately closed and the workers treated with antibiotics. No cases of anthrax were observed to occur among any persons who received antibiotics prophylactically.
The cases of inhalational anthrax can be grouped into three clusters defined by the geographical point of exposure: two cases in Florida, who were exposed to a contaminated letter that passed through the Florida media publishing company; four cases among the postal workers in Washington, DC, who were exposed to the Daschle letter at the Brentwood postal facility; and two cases among postal workers in New Jersey, who were exposed to the Daschle letter at the Hamilton postal facility. …