Reply to Osterholm and Kelley: H5N1 in humans—the threat is not as bad as it may seem

We are responding to the letter submitted by Osterholm and Kelley (1) regarding a portion of our recent commentary “H5N1 influenza viruses: Facts, not fear” (2), which pointed out that mild H5N1 infections are currently not accounted for under the World Health Organization (WHO) H5N1 case/fatality definition.

First, the basic complaint that Osterholm and Kelley make (1), that “the PNAS Perspective was an incomplete analysis and that no conclusions can be drawn from the data presented” is invalid. There was, in fact, no analysis presented in our PNAS piece—we simply cited the 10 largest studies that are available related to seroprevalence in humans. Similarly, we made no claim to provide a complete summary of available data on this topic. Based on the literature documenting mild/asymptomatic H5N1 infections in humans, and based on a common-sense analysis of the WHO criteria for H5N1 case confirmation, we stand by our assertion that the true fatality rate for H5N1 infections in humans is likely orders of magnitude lower than the currently cited figure of more than 50% (3).

Of note, Osterholm and Kelley (1) use a quote from our commentary to suggest that we reported seroprevalence rates “from 0.2 to 5.6% (2)”—this quote is taken out of context and misrepresents the content of our work. In fact, 20% of the studies we cite reported a seroprevalence rate of 0%, and this is stated very clearly in the sentence from which they draw. The passage actually reads: “Of the 10 largest studies of which we are aware (N ≥ 500), which document seroevidence in humans for H5 infections, two studies report no seropositivity and the other eight report rates ranging from 0.2% to 5.6%; the studies are mostly conducted in rural areas where H5 infections have been previously documented (2)”.

Second, there is absolutely no reason to disregard studies from any of the documented H5N1 outbreaks, as Osterholm and Kelley (1) suggest regarding data from the 1997 Hong Kong outbreak. We firmly believe that all available data on H5N1 viruses in humans are equally relevant.

Third, the statement by Osterholm and Kelley that the National Science Advisory Board for Biosecurity “recommendation regarding the redaction of two submitted manuscripts to Science and Nature was related to the transmissibility of the viruses… not driven by their potential virulence” is clearly disingenuous. H5N1 viruses would be of no interest to the National Science Advisory Board for Biosecurity if not for the case fatality rate of more than 50% that is currently reported according to the WHO definition (which is based on very severe cases only).

The high virulence and transmissibility that were originally reported related to the laboratory-derived H5N1 virus strain sparked worldwide alarm. Recently, however, there was a “revision” of the data, which suggests that the mutated H5N1 virus transmits less efficiently than do regular human influenza viruses and that, when transmitted, the mutated virus does not cause disease in ferrets (4).

One can only hope that the facts around H5N1 avian influenza viruses—as they come out—will prevail over fear and fear-mongers in the future.