One of the biggest mysteries about the 1918 Spanish flu pandemic was why it killed so many young people. Nearly half of the dead—which numbered in tens of millions—were adults aged 20 to 40. World War I ended in the middle of the pandemic, and ultimately, more U.S. soldiers died from the flu than in combat.

Why was this flu so deadly to this particular group? The answer, according to work by the University of Arizona evolutionary biologist Michael Worobey, had to do with the flu viruses those victims encountered as children, decades before. As dominant flu strains change over time, people born after 1889 had never encountered a strain similar to the Spanish flu, leaving them vulnerable in the pandemic.

This pattern that led to the deadliness of the Spanish Flu among the young was not a one-time fluke. A new study in Science from Worobey and his colleagues looks into two types of recent avian flu, H5N1 and H7N9, and again finds that the first type of flu a person was exposed as a child predicts which ones they will be vulnerable to in the future. In rosier terms, this means exposure to a certain human flu virus as a child could protect you from an only-somewhat-related avian flu virus.

“This a break from previous thinking that a new flu virus emerges into a completely blank state with no immunity in the population at all,” says Worobey. And it could mean different strategies for dealing with flu epidemics in the future.

Worobey and his team analyzed the number of people H5N1 and H7N9 made sick or killed in six countries with outbreaks between 1997 to 2015. Both viruses had jumped over from birds, so no one should have been exposed to them before. But what they found was stark: H7N9 affected many more people born before 1968, and H5N1 disproportionately affected young people born after 1968. “The data is really striking. The trends are really strong,” says Matthew Miller, a biochemist at McMaster University, who was not involved in the study. You can see for yourself.

H7N9 (red) and H5N1 (blue) cases and death by birth year of patient.     
Reprinted with permission from KM Gostic et al., Science 354:721 (2016)

What happened in 1968? Influenza is unique among viruses in that it comes in many, many different types, and in 1968, the dominant type of flu switched over from H1N1 to H3N2. If you’re getting dizzy from all the letters and numbers, don’t worry. What matters is that everyone born before 1968 was probably first exposed to a flu virus that belonged to one group of viruses (which scientists actually do just call “group 1”), and everyone born after 1968 was probably first exposed to a second group (“group 2”). And that provoked a lifelong immune reaction to all viruses in that group. It includes, even, a flu virus that usually sickens birds but is somewhat related to the one first sickened you. Your immune system imprints on the first flu virus you ever encounter.

The CDC currently recommends that everyone gets flu vaccines, but especially the elderly. The imprinting phenomenon suggests we might want to rethink who gets vaccines in a pandemic.“Even in the best case scenario, there are going to be limits on the availability of vaccines if a pandemic were to arise,” says Miller. “This would help you make some logical assumptions about who the most at risk groups are.” And those would be the people who did not encounter that particular group of flu virus as kid.

The finding could also influence the search for a universal flu vaccine. It’s unclear exactly why the immune system imprints on the first group of flu virus it encounters, but scientists do know some differences between the two groups. What makes them distinct is hemagglutinin, a protein that sticks out lollipop-like from the outside the virus. Usually, flu vaccines and immune systems in adults hone in on the candy part of the lollipop protein, which is very distinct and comes in 18 different flavors, from H1 to H18. All 18 of these types can be sorted into group 1 and group 2 though, based on the stick part of the protein. It could be that the immune system imprints on the first hemagglutinin stick it sees, so universal flu vaccine efforts should also focus on the stick part of the protein.

Ultimately, these patterns around bird flu as well as the Spanish flu point to the importance of understanding not just the biology of the virus itself but the people that it affects. Viruses are not intrinsically deadly so much as they are deadly when they meet the right host. The study of influenza, says Steven Riley, an infectious-disease researcher at Imperial College, is “the study of things at play over years, of interactions between viruses and long-lived human hosts over fifty, sixty, or seventy years.” Our bodies are reservoirs of our viral history.