This year the world will mark the 100th anniversary of one of the most devastating infectious disease events in recorded history: the 1918 influenza pandemic, which caused an estimated 50 million to 100 million deaths worldwide.

There were several reasons for the awful toll. First, most people likely had no preexisting immune protection to the brand-new strain that had emerged. Second, this particular virus may have been unusually lethal. Third, crowding and poor sanitation allowed for rampant disease transmission, especially in regions where access to health care was limited. And finally, antiviral drugs and flu vaccines were still decades in the future.

Over the past century we have made substantial advances in all these areas. But we are still unprepared for the inevitable appearance of a virus like the one that struck a century ago. Even an ordinary seasonal flu epidemic will still kill some 12,000 to 56,000 people every year in the U.S. alone. That is because seasonal viruses continually evolve, and although we update our vaccines frequently, they may be only 40 to 60 percent effective. Moreover, seasonal vaccines may provide little or no protection against pandemic flu. Pandemic viruses typically arise from a process referred to as an antigenic shift, in which the new virus acquires, usually from animal influenza viruses, one or more genes that are entirely novel (as seems to have happened in 1918, when all eight pandemic virus genes were novel).

In the years since 1918, three influenza pandemics associated with antigenic shifts occurred: in 1957, 1968 and 2009. In each of these instances, however, the new viruses emerged via the mixing of animal influenza virus genes with those of the 1918-descended viruses already circulating in the human population, which meant that many people were at least partially immune. That, plus lower viral pathogenicity and improvements in public health infrastructure and medical treatment, is what probably led to less catastrophic pandemics.

We must also tackle the issue of “prepandemic” influenza viruses—those that could potentially cause pandemics but that have not (yet) done so. Human infections with avian influenza viruses have occurred with increasing frequency over the past two decades. Prepandemic vaccines against various strains of H5N1 and H7N9 viruses have been developed and in some cases stockpiled; similar to seasonal influenza viruses, however, these avian strains are subject to antigenic drift within their avian hosts. Many of the H7N9 avian viruses that have jumped species from poultry to cause human infections in China in 2016 and 2017 have changed significantly from 2013 avian strains. As a result, the human immune responses elicited by a vaccine developed against the 2013 H7N9 virus may not be effective against 2017 strains.

The remarkable capacity of influenza viruses to undergo antigenic drift or shift to overcome and escape human population immunity leaves us vulnerable to a public health disaster potentially as serious as the 1918 pandemic. To meet this global health challenge, scientists are working to develop “universal influenza vaccines”—new types of inoculations that can provide protection not only against changing seasonal influenza viruses but also against the inevitable pandemic viruses that will emerge in the future.

Recently the National Institute of Allergy and Infectious Diseases convened a workshop with leading experts in the influenza field to address the need for better influenza vaccines. Among many obstacles to developing a universal vaccine, the most formidable is our incomplete understanding of the immune responses that protect people against influenza, including the role of immunity at mucosal surfaces.

One approach is to design a vaccine to generate antibody responses to parts of the virus that are common to all influenza strains and do not readily change by mutation. It is also crucial to clarify how other parts of the immune system work together with antibodies to protect against influenza. The hurdles in the development of such vaccines are daunting. But we are optimistic that we can apply existing tools and experimental strategies to meet the challenge. As we note the centennial of the 1918 flu pandemic, let us remind ourselves of the importance of this line of research in preventing a repeat of one of the most disastrous events in the history of global health.