February 23, 2012

expert reaction to meta-analysis of H5N1 surveillance data, as published in Science

The SMC sought comments in response to a meta-analysis in the journal Science which used blood serum data to show a much higher global rate of infection with the flu virus H5N1 than reported by the World Health Organisation (WHO).

Paul Hunter, Professor of Health Protection at the University of East Anglia , said:

“That many cases of H5N1 influenza are not diagnosed is not an especially surprising finding. Any surveillance system is only going to pick up a relatively small number of the cases circulating in the community. Furthermore, the sensitivity of surveillance systems varies considerably from disease to disease. In terms of influenza, we expect that most people who get infected will never receive a laboratory diagnosis – as this evidence confirms.

“One of the findings in this paper is that criteria for serological diagnosis on H5N1 as developed by WHO underestimate positive rates compared to those derived by other workers. Choosing criteria for disease diagnosis is complicated – just because you detect more positive cases doesn’t mean your criteria are better. A threshold that comes up with more positives will include more false positives; one that detects fewer positives will miss some cases. This is an interesting contribution to the discussion of how H5N1 should be documented, but I’m not sure it identifies a burning problem or an alarming level of under-reporting.”

Dr Andrew Hayward, Reader in Infectious Disease Epidemiology, University College London, said:

“This important study illustrates the critical importance of understanding community levels of infection when assessing the severity of novel strains of influenza. The cases that are most likely to be tested for H5N1 are severe cases that end up in hospital. People who are ill enough to get tested for H5N1 have a very high risk of dying but this work suggests that many more people in the community become infected. We already know that many infections with seasonal influenza strains or with the recent H1N1 pandemic have no illness or only a minor illness. This work suggests the same may be true for H5N1 which we have previously assumed to be very rare but with a very high death rate. As mild cases do not seek medical attention they are difficult to count. International systems to count these ‘silent’ cases are needed to understand the true severity of H5N1 and allow us to gauge appropriate levels of response.”

‘Seroevidence for H5N1 Influenza Infections in Humans: Meta-Analysis’ by Taia Wang et al., published online in Science on Thursday 23rd February