“Confidentiality of sensitive national outbreak surveillance data assured!” This prominent guarantee on the website of the South East Asian Nations Infectious Diseases Outbreak Surveillance Network says it all. Open sharing of data often ends when it could compromise trade or other national interests.

Massive public databases exist in many areas of science, and are critical to cutting-edge research. But there is no comprehensive database of outbreaks of infectious diseases. We have better data on galaxies 10 billion light years away than on human cases of avian flu in China or Vietnam. Yet the world is imperilled by outbreaks, wherever they happen.

It is difficult enough to gather the data in the first place. At the moment, plague, cholera and yellow fever are the only notifiable diseases that countries must report to the World Health Organization (WHO). Vietnam has often reported human cases of avian flu months after the event, and outbreaks in animals have been concealed in many countries.

Not before time, the WHO will have broader powers in 2007, when international health regulations, agreed by its members in May last year, come into force. These impose obligations on states to respond to any infectious disease of international concern. Cat-and-mouse games will no doubt continue, but the WHO will at least have a ‘health policing’ role, something that it currently lacks.

The reporting of avian-flu cases has recently improved in speed and openness, but the quality of the available data remains dire, and biological samples are insufficient. A typical WHO update will give the number of new cases from a country and, on a good day, the age, sex and rough location of each case. But there is little information on familial case clusters, and typically no clinical data. What is worse, these few data are in text form strewn across hundreds of individual WHO web pages.

The world badly needs a database for outbreaks of avian flu.

Data on outbreaks in poultry are even more sparse, and mostly come from the World Organisation for Animal Health (OIE). Someone at the UN Food and Agriculture Organization (FAO) is maintaining a file of cumulative bird outbreaks from OIE and other data, and is making it available to researchers and journalists. But it is incomplete, lacks good location data and contains errors.

Genetic data are also lacking. When samples are sequenced, the results are usually either restricted by governments or kept private to an old-boy network of researchers linked to the WHO, the US Centers for Disease Control and Prevention, and the FAO. This is a far cry from the Human Genome Project, in which all the data were placed in the public domain 24 hours after sequencing. Many scientists and organizations are also hoarding sequence data, often for years, so they can be the first to publish in academic journals. With the world facing a possible pandemic, such practices are wholly unacceptable. Nature and its associated journals are not alone in supporting the rapid prior exposure of data when there are acute public-health necessities.

Three cheers, then, to Ilaria Capua of the Tri-Veneto Region Experimental Animal Health Care Institute in Italy, who last month threw down the gauntlet to her colleagues by refusing to put her latest data on Nigeria and Italy in these private networks. Instead she uploaded them to GenBank and called on her colleagues worldwide to do likewise. Only in this way can researchers establish and track the global pattern of the evolution of the bird-flu virus.

Imagine scientists anywhere being able to log on to a publicly available, searchable Internet database, updated in real time, with full clinical and sequence data on each human case, and accurate and complete poultry data. Dream on. The WHO's clunky online Global Health Atlas, which gives rough aggregate data for many diseases, doesn't have a category for H5N1 under influenza. The Global Infectious Diseases Epidemiology Online Network (GIDEON) database contains only the data on avian flu that it extracts from the WHO's updates and reports from the ProMED reporting system for infectious diseases. ProMED itself has pioneered outbreak alerting, notably during the SARS crisis, but its content consists largely of media cuttings. Such aggregation is often done faster and better by bloggers, and by the Global Public Health Intelligence Network (GPHIN), a Canadian intelligence operation that provides an early warning system that screens media and blogs in seven languages in real time.

The world badly needs a database for outbreaks of avian flu. But international agencies are failing to rise to the challenge.