To slow the spread of highly pathogenic avian influenza — such as the prevalent H5N1 strain — real-time reporting of current genome data is crucial. Yet “extensive delays” of 7.5 months or more between sample collection and submission to the Global Initiative on Sharing All Influenza Data (GISAID) repository have occurred, according to a commentary by Canadian experts.
Early in the COVID-19 pandemic, for instance, a study highlighted lengthy delays in severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) sequence submission to GISAID, averaging about 48 days between sample collection and submission. After the analysis was published, numerous countries improved their pipelines for data submission, bringing the timeline down to 30 days by 2024.
Improving the collection-to-submission times worldwide helped with SARS-CoV-2 variant monitoring and public health recommendations in recent years, the authors wrote. The same efforts are needed to track and mitigate H5N1.
“We initially went to look at how H5N1 was evolving within Canada and were surprised to find that we couldn’t tell because so few of the sequences were recent,” lead author Sarah Otto, PhD, professor of zoology at the University of British Columbia’s Biodiversity Research Centre in Vancouver, told Medscape Medical News.
“There is very little targeted funding devoted to monitoring H5N1 in wildlife in North America, but evidence suggests that wildlife, particularly migrating birds, is the source of infections getting into poultry and cattle and from farm animals into humans,” she said. “We need more targeted and rapid sequencing of wildlife to detect early signs of infections in a region and proactively protect local farms and farmworkers.”
The commentary was published online on March 25 in Nature Biotechnology.
Analyzing GISAID Data
Otto and co-author Sean Edgerton, a PhD student focused on genomics and infectious diseases at the University of British Columbia, accessed GISAID data on January 10 and February 26. Their analyses included sequences for H5N1, H1N1, and SARS-CoV-2.
For H5N1, the researchers analyzed metadata for clade 2.3.4.4b sequences submitted from January 2021 through December 2024, removing data collected more than 1 year before this period. This analysis resulted in 18,817 samples from 84 countries.
Overall, the median collection-to-submission time delay was 228 days for H5N1 data, ranging from 14 to 932 days. Among countries submitting at least 50 sequences, median delays ranged from 25 in the Czech Republic and the Netherlands to 618 days in Canada. Half of the countries had median delays longer than 6 months.
In general, influenza metadata on GISAID appeared to be vague or missing in most cases, Otto and Edgerton wrote. About two thirds of submissions lacked sufficient detail about the host species to determine the genus of the host, and 90% lacked information about whether the animal was domestic or wild. In addition, 87% lacked animal health status information.
Certain data sources, particularly the US Department of Agriculture (USDA), provided incomplete information when first submitting sequences and added details later, the authors wrote. When accessing GISAID in January and February, the authors saw the backfilling of metadata, where the USDA later added more information about sample collection day or state for about half of 1763 sequences.
Current data delays were similar to those during the H1N1 pandemic a decade ago, the authors wrote. For the 21,387 samples from 150 countries submitted from 2010 through 2013, the median collection-to-submission time delay was 289 days, ranging from 32 to 1115 days.
Otto and Edgerton considered several factors that might account for delays, such as gross domestic product (GDP). However, several high-GDP countries had long delays, including 413 days in Norway, 467 days in France, and 618 days in Canada, while several lower-GDP countries submitted data two or three times faster.
One factor — the host species — appeared to be associated with a longer delay, the authors wrote. For instance, with H5N1, the collection-to-submission time was only 15 days for the first 65 samples sequenced from humans compared with 230 days for 18,592 samples from mammals or birds. Wildlife samples took even longer.
Similar delays occurred with H1N1 samples, the authors found. The median collection-to-submission time was 244 days for 17,600 human samples compared with 422 days for 3782 animal samples. Even in 2024, H1N1 data had a median of 60 days for 35,681 human samples compared with 127 days for 922 animal samples.
Animal samples also had longer delays for SARS-CoV-2 data. In 2024, the median delay was 30 days for 693,824 human samples compared with 787 days for 654 animal samples.
The differences could be attributed to data collection and the agencies that handle animal sequencing, the authors wrote. The USDA or Canadian Food Inspection Agency, for instance, may have certain policies for sample processing and data sharing for animal sequences, especially when multiple groups and agencies may be involved in data collection.
Although submission delays have improved during the past decade, shifting from a median of 14 months for H1N1 data to about 7.5 months for H5N1 data, the ongoing slowdown in submissions can prevent real-time analysis of “what is happening on the ground,” the authors wrote.
“Viruses, particularly this virus [H5N1], do not know boundaries, either between countries or species,” Otto said. “While we’ve been lucky so far that the virus hasn’t yet evolved or spread readily among humans, rapid monitoring of wildlife is needed to know when the virus is getting close so that we can work together to contain the virus.”
Looking Ahead
Canadian virologists and wildlife experts have been keeping a careful eye on H5N1 developments in recent years, looking for mutations and the potential for human-to-human transmission.
“The scientific community has been watching the H5N1 viruses mutate and evolve in avian populations around the world for several decades. In the past few years, these viruses have shown increasing flexibility in infecting diverse mammalian populations, which raises their pandemic potential significantly,” said Brian Ward, MD, professor of medicine at McGill University and associate director of McGill’s J.D. MacLean Centre for Tropical Diseases in Montreal.
“Timely monitoring of H5N1 mutations — particularly those that increase the risk of mammal-to-mammal transmission — is important to give us as much forewarning as possible of an impending pandemic,” he said. “Even a few months could make a large difference in how effectively governments, industry, and populations can respond. International cooperation in keeping the GISAID database current is a critical tool in this effort.”
Worldwide surveillance remains key — and is growing more important — Canadian experts told Medscape Medical News, both with official sample sequencing and informal monitoring. Healthcare providers can remain alert to patients who present with unexplained severe respiratory symptoms, they said.
“Ultimately, from a human health standpoint, I’m not directly concerned about the circulating H5N1 strains. I’m worried about what they might become,” said Scott Weese, DVM, professor of pathobiology at the Ontario Veterinary College and director of the University of Guelph’s Centre for Public Health and Zoonoses in Guelph, Ontario.
“H5N1 is currently not well adapted to infect and transmit between people and most domestic animals. Severe disease has been rare in people,” he said. “However, the concern is what this virus could become as it continues to evolve, and the more it infects mammals, the greater chance for relevant evolution or recombination with other flu viruses.”
That’s why making surveillance-to-action plans is key for H5N1 and other emerging diseases, Weese said. Real-time data help public health workers, government leaders, and healthcare providers know what to do. With months-long delays, however, taking certain preventive actions would likely be too late, he noted.
“Cats should show us why we need to be diligent. This virus effectively infects and often kills cats. If it did to us what it does to cats, the world would be very different,” Weese said. “It’s not currently milder in people than cats because of anything we did or any predictable process. It’s luck. This virus is a great match for cats — and readily infects their brains. We don’t want a version that can do the same to people.”
The analysis was conducted without outside funding. Otto, Edgerton, Ward, and Weese reported having no relevant financial relationships.
Carolyn Crist is a health and medical journalist who reports on the latest studies for Medscape Medical News, MDedge, and WebMD.
