In an ideal experimental world, to understand long COVID’s risks, researchers would systematically survey large swaths of the population over long periods of time, watching to see who gets infected, who goes on to develop the condition, what form it takes, and how it impacts people’s health, says Shruti Mehta, an infectious-disease epidemiologist at Johns Hopkins University who is studying long COVID. But few institutions have the resources for such an undertaking, which could span many months or years. So many researchers have to make do with the limited data sets that are already available to them. As a result, some studies end up biased toward patients who were hospitalized, while others wind up favoring people who have the time, means, and trust in the health-care system to sign up for long-term studies. Neither group fully captures long COVID’s wide-ranging toll. The situation’s especially tough for pediatric patients, who might be too young to articulate the severity of their symptoms and are often excluded from long-COVID studies. Long COVID certainly exists in kids, but it may not perfectly mirror what goes on in adults: Children’s susceptibility to the virus is different, and their bodies are so rapidly changing, says Yonts, who runs a pediatric-long-COVID clinic in D.C.

All told, the study of long COVID has become, as Sanders of MIT puts it, “a data disaster.” Some researchers estimate that a single-digit percentage of SARS-CoV-2 infections bloom into long COVID; Al-Aly is one of them. Others, meanwhile, favor larger numbers, with a few even insisting that the rates are actually more than half. Most of the experts I spoke with said they feel comfortable working in the 10 to 30 percent range, which is where many studies seem to be starting to converge. Finding one answer is tricky, without knowing how many forms long COVID can take—some could be more common than others. Formally splitting the disease into subdivisions could help address some of these ambiguities. But we don’t know nearly enough to start slicing and dicing, says Bryan Lau, an infectious-disease epidemiologist working with Mehta and Priya Duggal.