Last week, a North Carolina man became a notorious microbial killer’s first confirmed victim this year. Fifty-nine-year-old Eddie Gray had unknowingly come across a brain-eating amoeba while swimming in a man-made lake near Fayetteville in mid-July; 10 days later, he was dead.

Since the brain-eating amoeba was first recognized and named, in 1970, grisly reports of its disastrous attacks have made headlines nearly every year. About 97 percent of confirmed cases in the United States have been fatal. But the infection is also incredibly rare, and the small sample size leaves the epidemiologists who study it and the doctors who encounter it with their hands tied. It may be one of nature’s most perfect crimes.

Despite their gruesome moniker, most brain-eating amoebas never eat a single brain. The single-celled swimmer, formally known as Naegleria fowleri, passes its time resting in a dormant state or, when it’s warm enough, splashing around and munching on bacteria. Unlike most waterborne pathogens, it’s utterly benign if you drink it. It becomes dangerous only when, thanks to a person enjoying a day at a water park or a quick rinse in a stream, the amoeba is yanked from its bacterial buffet and swept into the dark recesses of the human nose.

If the amoeba isn’t already in dining mode at this point, the intense burst of human body heat can help it shape-shift out of dormancy. Like anyone else waking up somewhere unfamiliar, the brain-eating amoeba is desperate for a food source, so it slithers its way up the olfactory nerve until it spies a tasty-looking tangle of neurons and digs in. The host’s immune system, sensing an unwelcome visitor, sends an onslaught of white blood cells to take down the feasting parasite. That commotion leads to a swollen—and, eventually, irreparably damaged—brain.

“We have these hard skulls to protect our brains from traumatic injuries,” says Jennifer Cope, a medical epidemiologist who studies N. fowleri for the Centers for Disease Control and Prevention. “But when you start to swell, that hard skull actually gets in the way.” With nowhere else to go, the brain stem and other nearby regions are pushed down through the bottom of the skull. For most of the brain-eating amoeba’s victims, this crowding is the direct cause of death.

Altogether, this deadly infection is known as primary amebic meningoencephalitis (PAM). It’s similar to viral and bacterial meningitis, except the invader that causes it comes from the water and eats your brain. With just about 30 cases in the United States in the past decade, PAM is an uncommon variant of an already rare disease. Since N. fowleri was first identified by Australian doctors roughly 50 years ago, there have been at least 146 confirmed cases of PAM in the United States, with only four survivors.

N. fowleri is so alarmingly dangerous because, like bacterial and viral meningitis, PAM is nearly impossible to spot until long after any of the interventions that have worked for some victims are even an option. Even when a patient does make it to a hospital within a few days of infection, PAM is often mistaken for one of these types of meningitis (distinguishing them requires an invasive spinal tap) and treated as such, to no avail. Experts agree that for every confirmed case, there are likely one or two that were misdiagnosed and recorded instead as more common meningitis fatalities.

Gray’s death fits the most common narrative of an N. fowleri fatality. Though not much is known about the organism, Cope says, “we do know it’s thermophilic—it likes heat.” Swimmers in balmy southern lakes and rivers have generally been the unlucky few each summer, but in recent years there have been some concerning outliers.

In 2013, a 4-year-old boy living near New Orleans died unexpectedly from what doctors later determined to be PAM. The CDC was called. “I asked the same questions we usually do about swimming in lakes, and [the parents] said they hadn’t been to any lakes,” Cope says. “It led to us finding that really the only thing was that he’d been playing on the backyard Slip ’n Slide—with a hose hooked up to the backyard tap.”

Two years earlier, a 20-year-old man in the same parish had developed PAM after rinsing his sinuses with a neti pot he’d filled up in the sink. “We went back and we did sampling again,” Cope says. “This time we found it in the home where the patient had been exposed, but we also found it in the drinking-water distribution system, in the pipes that led into the home from the street.” It was the first time the amoeba had been spotted in a centrally treated U.S. system.

“What the problem was in Louisiana was that they were using chloramines,” says Charles Gerba, a microbiologist at the University of Arizona who studies groundwater N. fowleri. In Arizona, Gerba says, chlorine is the standard choice for disinfecting municipal water, but about 30 percent of the country instead uses chloramine, a mix of chlorine and ammonia that leaves behind fewer harmful by-products such as chloroform. Chlorine and chloramine dissipate differently in water, meaning that under the right set of circumstances—for instance, low water turnover in areas left partially vacant by Hurricane Katrina—the less aggressive chloramine can leave some pockets untreated.

The contaminated system in Louisiana was flushed in 2013, and the Environmental Protection Agency began regularly monitoring water in the region. But the process is complicated, and spotting the amoebas can take weeks once they show up. There’s no quick test for the parasite, in part because so much about it remains unknown. Cope’s lab is currently in the early stages of sequencing the genomes of N. fowleri samples that the CDC has collected from across the United States over the years. It’s identified three genotypes so far, she says, but the system “doesn’t have a lot of granularity to it” and isn’t used much yet. Its goal is to eventually develop a sort of amoeba dictionary detailed enough to match samples from patients to specific collections of the creatures, giving scientists the ability to pinpoint exactly where someone encountered the pathogen.

Today the clearest risk factor is still warm water temperatures, which make it easier for the microbes to reproduce. “What’s concerning people,” Gerba says, “is that with global warming, the water is warming up, and people are expecting to see more cases.” Brain-eating amoebas around the world are generally relegated to sunny climes, but when two PAM cases were confirmed in Minnesota in 2010 and 2012, it raised the specter of a climate-change-induced N. fowleri explosion. In a 2017 review, Cope and her colleagues at the CDC warned that the Minnesota cases could be a preview of things to come. Today she emphasizes that despite any geographical changes, the number of annual cases in the United States remains steady, and too small to test for any statistically significant trends.

Given how rare an N. fowleri infection is, worrying about it is like worrying about falling into a volcano: The probability of it happening to you is small, and making it even smaller is simple. All it takes is steps such as purifying tap water an extra time before flushing your nostrils with a neti pot and not submerging your head (or using a nose clip) when you swim in nature—even if the water looks really, really clean.

“I worry about the word natural all the time. Every time you swim in natural waters, you’re at an increased risk of getting ill,” Gerba says. “People think, Oh, it’s so natural; it’s so fresh, and I go, ‘Yeah, all the birds pooped in it this morning.’”