As decision-makers weigh whether to reopen U.S. schools this fall, they face a daunting reality: COVID-19 appears to have surged among American kids this summer. On July 9, there were approximately 200,000 confirmed cases in children; by Aug. 6, that number rose by nearly 180,000, an approximately 90% increase in just over four weeks, according to a new report. That translates to an increase from 319 per 100,000 American children to 501 cases per 100,000, in just four weeks.

The report, from the American Academy of Pediatrics and the Children’s Hospital Association, counts more than 380,000 total confirmed cases in children as of Aug. 6.

Recently, evidence has emerged that COVID-19 seems to not to produce as severe symptoms in children as it does in adults, which has given new justification for communities and lawmakers to push to reopen schools for in-person classes. However, these new data, compiled by the American Academy of Pediatrics and the Children’s Hospital Association, are a warning sign that reopening schools without proper social distancing measures in place could accelerate the spread of COVID-19. This is especially concerning for middle and high schools, because older kids with COVID-19 are thought to spread the virus at rates similar to adults. And while severe COVID-19 in children is uncommon, it’s not unheard of. Children made up 0.5% to 5.3% of all hospitalizations in 20 states (plus New York City) that reported figures, and 0.3% to 8.9% of all kids who got sick were hospitalized. As of Aug. 6, 90 children had died from the disease, in 44 states and NYC.

Overall, the new report found that children make up a growing proportion of COVID-19 cases. Back on April 16, kids only made up 2% of total COVID-19 cases (in 46 states, plus NYC and D.C.); by Aug. 6, they made up about 9.1% of cases nationwide (in 49 states, plus NYC, D.C., Puerto Rico and Guam).

These data do not paint a perfect picture of how COVID-19 is affecting children in the U.S. Each state has rolled out social distancing measures and COVID-19 testing differently, which might affect both who’s getting sick and who’s being counted. It’s also unclear to what extent increased testing has led more children to be tested for the virus in the first place. It might be the case that in the early months of the pandemic, only the sickest were getting tested; given that children seem to experience less severe symptoms of a COVID-19 infection, it’s possible that in the spring, many children caught COVID-19 and recovered without ever getting noticeable symptoms, and thus without ever getting tested.

States also have different methods for collecting and organizing data about COVID-19. Notably, the age range that defines “child” differs between states—while most reported ages 0 to 17 or 0 to 19 as children, Florida and Utah limited “children” to ages 0 to 14, while Alabama went all the way up to age 24. These differences could have a big impact, because older children seem to be more likely to have symptoms than younger ones, and are more contagious.

However, the report does make certain trends clear. States in the south and west have driven much of the increase in the number of children diagnosed with COVID-19. Alabama had the highest proportion of cases in children, with kids making up 22.5% of the states total confirmed as of Aug. 6, followed by Wyoming and Tennessee, at 16.5% and 16.2%, respectively. Meanwhile, New York City and New Jersey, the epicenters of the early days of pandemic, have a much smaller relative proportion of children diagnosed with COVID-19, comprising about 3% in each.