Anthrax: Filling in the Blanks

Almost everything we know today about anthrax outbreaks has been inferred from a 1979 epidemic. In Sverdlovsk, Russia, a germ warfare facility accidentally released an aerosol of the pathogen, killing scores of people. Casualties were relatively light because the wind had blown the anthrax away from Sverdlovsk, toward the countryside. (One Russian bioweapons expert has estimated that if the wind had been blowing toward the city that day, hundreds of thousands would have died.)

But the Sverdlovsk data are sketchy because the Russians waited 13 years before allowing in a U.S. investigative team (which included Alexander Langmuir, MD, MPH '40).

Last year, D.A. Henderson, MD, MPH '60, then-director of the School's Center for Civilian Biodefense Studies, decided to lay out all that was known about the Sverdlovsk incident for Ronald Brookmeyer, PhD, an expert in using biostatistics to monitor and track epidemics. He hoped Brookmeyer could wring some additional information from the skimpy facts:

• The spores were leaked April 2, 1979.
• The Russians had mounted a tardy public health response by mid-month. 
• At least 70 people died. 
• The dates of death of all known victims were learned through photographs of their tombstones. 
• Some dates of the onset of symptoms were known.

Brookmeyer, a professor of Biostatistics, wasn't fazed by the dearth of data. "Ideally," he now says, "you of course want good surveillance data, but that's the exception rather than the rule. Often you have to settle for incomplete numbers—and that's when biostatistics can sometimes fill in some blanks." 

The investigators decided to work from two main assumptions: that all the deadly spores had originated with the April 2 leak, and that, over time, the anthrax cases would have naturally distributed themselves along a log-normal curve (a curve with a long right tail that signifies the cases with long incubation periods). 

Tweaking the data, they hit pay dirt. The scientists' model showed the disease's average incubation period—that is, the median time for onset of inhalation anthrax—was 11 days after exposure: an incubation period nearly twice as long as any previous estimate. Further, they were able to show that 1 percent of those exposed on April 2 could have fallen ill as long as 58 days after the accident. 

Brookmeyer had confirmed that, in any future anthrax outbreak, antibiotics would have to be kept on hand for a much longer time than hitherto imagined, and that those exposed to the pathogen should continue to receive treatment for 60 days.