UPTON SINCLAIR’S NOVEL THE JUNGLE ARRIVED IN BOOKSTORES IN 1906, and the public outcry it unleashed was swift. Sinclair had spent seven weeks undercover in the Chicago meatpacking district, and his tales of the unsafe labor conditions he found there were intended as a call for labor reform. Instead, readers fixated on his stomach-turning descriptions of the industry: dead rats thrown in with the food, abscessed cow meat and contagious diseases on the processing line.

Worries about food safety even reached an international audience. (A young Winston Churchill wrote one of the book’s early reviews in the United Kingdom.) In the United States, President Theodore Roosevelt, who had received a personal copy from Sinclair, was appalled by what Sinclair had unearthed. Roosevelt wrote back, saying that “the specific evils you point out shall, if their existence be proved, and if I have power, be eradicated.”

President Roosevelt indeed championed two pieces of sweeping food safety legislation, both passed within six months of the book’s publication. The federal body set to enforce the food and drug laws was eventually named the Food and Drug Administration. And while the FDA today may be known mostly for its role in approving new drugs and medical devices, food safety still commands by far the largest single portion of the agency’s time and budget.

Since The Jungle, the agency’s record on food has largely been one of adapting to new threats, such as determining acceptable levels of chemical contaminants in fish, and finding ways to head off food bioterrorism. And food-borne illnesses have gone down steadily under its watch, dipping 22% in the period from 1996 to 2012 for six major pathogens.

But then something happened—or rather, nothing happened. Little to no progress has been made in reducing the incidence of food-borne illnesses since 2012, according to an April report from the Centers for Disease Control and Prevention. And in a frustrating twist, the reason for that standoff may be a new way that doctors test for these illnesses.

Until recently, when people showed signs of having eaten tainted food, a stool sample was cultured at the hospital. That culture was sent to a local health department to identify the strain of the pathogen. This data was then uploaded to PulseNet, a national network of labs coordinated by the CDC, which gathers and shares information on food-borne illness and specific pathogens in an attempt to detect outbreaks. If PulseNet found a match between a specific pathogen strain in multiple stool samples, the infection could then be traced to a common source.

In 1992 and 1993, for instance, a specific strain of the Escherichia coli bacterium, O157:H7, swept four western states, infecting some 600 people and killing four. Health officials collected stool samples, interviewed the patients, and used the signature of the pathogens they found to help trace the source to a single chain of restaurants that had been serving the same infected meat across a wide part of the country, and eventually to six suspect slaughter plants. Public health experts were able to step in and solve the problem.

But these culture-based samples, as useful as they are, take days to grow. More recently, many hospitals have instead adopted CIDT, a rapid swab test that can identify the pathogen based on specific antigens or genetic sequences of germs. While the CIDT is quick, it doesn’t characterize specific strains of pathogen. Without that information, pathogen signatures don’t get uploaded to PulseNet, which then has more difficulty linking multiple cases of food-borne illnesses in the search for connections—one possible reason that food-borne illness rates haven’t moved much in recent years.

The next generation of monitoring may help. The FDA and CDC now identify food pathogens by sequencing their whole genomes, a method that can identify bacterial strains with even more precision and detail than the one PulseNet has used. Since 2012, scientists at an FDA-managed network of labs across the country, known as GenomeTrakr, have been collecting samples of pathogens, sequencing their genomes and uploading their genetic data into a new database housed at the National Center for Biotechnology Information. In 2014, FDA samplers uploaded a strain of Salmonella Braenderup from a nut butter plant in Oregon to GenomeTrakr, and found a match with a strain of salmonella found in samples from a small number of patients who had been sickened in the previous months. The FDA quickly warned consumers, got products recalled, and was able to stop the outbreak before it advanced.

But this system requires that physicians collect cultures from infected patients. So the CDC is exploring other ways to get its hands on that data about pathogens, both by encouraging labs to culture specimens with positive CIDT results and by working directly with CIDT manufacturers to create new kinds of tests that might collect more specific information about the pathogen, or keep some of the specimen organisms alive so that public health labs could culture and test them.

No single solution has nudged the needle yet. But the agencies hope that within the next few years, the steady elimination of food-borne illnesses will continue its century-long march.