REPLY
We are gratified that Ron Fouchier has joined (1) the important effort to quantify the risks (2–5) of the creation of potential pandemic pathogens, including ferret-transmissible variants of influenza A/H5N1. However, we disagree with many aspects of his assessment.
As in our article and Fouchier’s letter, here we proceed through the calculation, starting with probability of laboratory-acquired infections and the conditional probability of sparking a pandemic given such an infection and concluding with the consequences thereof. We then discuss some more general considerations.
PROBABILITY OF A LABORATORY-ACQUIRED INFECTION (LAI) IN A LAB WORKING ON PATHOGENS WITH PANDEMIC POTENTIAL
Fouchier bases his calculations on one of the sources we also used, the tabulation by Henkel et al. of reports of accidents involving select agents in the United States between 2004 and 2010 (6). However, he argues that the risk in his laboratory is considerably lower than the lower bound obtained from these reports of 0.2% per laboratory-year in a biosafety level 3 (BSL3) laboratory. He states, “These estimates, however, do not take into account specific pathogen types or research settings. This is crucial, because working practices in, e.g., virology and microbiology laboratories are different and because each biosafety laboratory is unique” (1). He proposes an alternative calculation based on 0 viral laboratory-acquired infections (LAI) in BSL3 labs over 2,044 lab-years in BSL2, -3, and -4 labs with select agents (6) and suggests that the proper value is <1/2,044 lab-years, or <5 × 10−4/lab-year.
These numbers are both conceptually and statistically invalid. While bacteriology and virology labs certainly perform some different activities, neither the references cited by Fouchier nor any other evidence of …
