Antibody analysis of blood samples is the only way to accurately track the evolving flu pandemic. Credit: C. Anand/AP

Researchers are turning their attention to one of the great unknowns about the ongoing H1N1 influenza pandemic: how many people have been, and are being, infected. The first surveys to monitor for antibodies to the virus are now getting under way, belatedly in some countries such as the United States. The findings could substantially change much of what epidemiologists know about the current pandemic.

"I'm very struck that we don't have even an idea of the magnitude of infection," says Xavier de Lamballerie, a virologist at the University of the Mediterranean Aix-Marseille II in Marseilles, France. "Epidemiologists haven't a clue if it is 5%, 10% or 20% of the population."

Gathering that information is crucial for improving estimates of pandemic spread, severity and mortality, and informing policies such as how to distribute vaccines and antiviral drugs.

Laboratory-confirmed cases of pandemic H1N1 underestimate the true prevalence by several orders of magnitude, as only a tiny fraction of cases can be tested. Instead, public-health agencies such as the US Centers for Disease Control and Prevention (CDC) in Atlanta, Georgia, use proxy measures, including data on the frequency of people reporting influenza-like symptoms to their doctors.

But testing blood samples for antibodies to pandemic H1N1 is the only definitive way to establish how many people have been exposed to the virus and to begin to estimate how this is changing over time. "Arguably, these data are one of the most important quantities," says Marc Lipsitch, an epidemiologist at the Harvard School of Public Health in Boston, Massachusetts, who is working with the CDC on pandemic flu.

Britain, France and Vietnam are among those farthest ahead with such studies for H1N1. In Britain, Andrew Hayward heads FluWatch, which for the pandemic has scaled up its seasonal-flu work to a £2.1-million (US$3.5-million) study co-funded by the Medical Research Council and the Wellcome Trust. Instead of its usual 650–850 subjects, the group will investigate 10,000 subjects, including 2,500 in a serology study to look for antibodies against H1N1 in their blood serum.

Arguably, these data are one of the most important quantities. ,

The UK Health Protection Agency has also launched a £180,000 study. This draws on blood samples collected from hospital patients for other purposes — so what it lacks in terms of targeting well-characterized groups of individuals, it makes up for in speed by using existing samples. Survey leaders have collected 1,403 blood samples from before the first pandemic wave, and 1,954 taken in August and September, across all age groups from eight regions in England. "I believe England is the first to obtain such seroincidence data," says Elizabeth Miller, the study's lead researcher.

She declined to comment on the results because they are under review at a journal, but says the data will provide "insight into the extent to which surveillance has underestimated the true burden of infection due to the occurrence of mild or asymptomatic infections". Hayward is also writing up preliminary FluWatch results gathered during Britain's first pandemic wave, and these too are likely to show a very different picture from that provided by surveillance data alone.

In France, the second wave of the pandemic has only just begun, buying precious time for the €300,000 (US$450,000) SéroGrippeHebdo ('Sero Flu Weekly') study, led by the French School of Public Health in Rennes and Paris. This project is recruiting 30,000 pregnant women, and gains speed by piggybacking on existing infrastructure for routine blood sampling of this group. It will publish serology data from 800 of the women weekly in real time, beginning this week. Already, says de Lamballerie, whose lab is doing the testing, "we've got a great baseline — no higher than 5% or so of the study population has already been infected".

In Vietnam, Peter Horby, a researcher at the National Institute for Infectious and Tropical Diseases in Hanoi, is part of a project testing leftover serum from haematology and biochemistry labs in nine provincial hospitals. He is also switching a seasonal-flu study of 908 people in 269 households in the northern Ha Nam province to study the serology of pandemic flu. Horby says it should yield good data on the true epidemiology of H1N1 in Vietnam.

Other projects are also just getting started. On top of the survey of pregnant women, Antoine Flahault, dean of the French School of Public Health, is seeking support for a French-led international study called CoPanFlu. This would see each partner fully profile 1,000 households during the pandemic, including serology testing at six-monthly intervals over two years. "One shouldn't underestimate the difficulty in getting these sorts of studies off the ground," says Hayward.

The United States was less prescient, it seems. Academic groups there are still in the process of applying for funds for such surveys, and Nature has also learned that the CDC is about to announce a serology study across ten states. During pandemic planning before the current virus arose, the United States extensively discussed the need for such studies but decisions weren't taken, says Donald Burke, dean of the Graduate School of Public Health at the University of Pittsburgh in Pennsylvania: "None of these very important studies were in place when we knew there was going to be a pandemic. It's unfortunate."