When George Badame started complaining about stomach pain after a big Italian Easter dinner three years ago, his wife, Mary, chalked it up to a little agita.
He vomited that night, and his stomach still hurt the next morning. The couple figured he had a virus, and Mary went to work as usual.
Back at their South Philadelphia home by midafternoon, she found George on the bedroom floor shivering.
Less than 24 hours after Easter dinner, he was in the intensive-care unit at the Hospital of the University of Pennsylvania. Doctors told Mary that her husband, 69, was in septic shock. His body's immune system had reacted so fiercely to an infection in his bloodstream that his organs were in peril.
"I looked at them perplexed," recalled Mary, 56. "I was so unaware. I didn't know a thing."
On June 13, just two months later, he died of the condition. And as in many sepsis cases, the cause of the terrible infection is unknown.
Every year, more than 1.6 million Americans - about one person every 20 seconds - are diagnosed with sepsis or septic shock, and 258,000 die, according to the Sepsis Alliance. That's more than die from prostate cancer, breast cancer, and AIDS combined, but other experts believe even that high figure is an understatement.
Sepsis is deadly because it is so difficult to diagnose in its early stages. Now researchers are working with data from millions of patients to tease apart the mystery of how to diagnose it sooner, before it turns deadly.
Some progress has been seen. While the number of sepsis cases is rising, the mortality rate is falling to a national rate of 15 percent to 20 percent, says Clifford S. Deutschman, a critical-care specialist who has studied sepsis for 30 years. But the mortality rate for patients such as George Badame who progress to septic shock is still more than 50 percent.
"We don't know enough about the underlying biology to be able to say, 'This is how septic shock is different from sepsis,' " Deutschman said.
In 2014, Deutschman cochaired a task force of 16 international experts from the Society of Critical Care Medicine in America and Canada and the European Society of Intensive Care Medicine with the goal of revealing the early symptoms of sepsis and distilling that information into a practical diagnosis tool.
The team took information from six million patients and came up with the Third Consensus Definitions for Sepsis and Septic Shock, published in February in the Journal of the American Medical Association.
"The new definition is that sepsis is life-threatening organ dysfunction caused by a disregulated host response to infection," Deutschman said.
The first attempt to identify the symptoms using clinical data was in 1991. A decade later, that work was updated and defined sepsis as a presumed infection with an inflammatory response that elevated two of four measures - heart rate, respiratory rate, white blood cell count, and temperature - or the Systemic Inflammatory Response Syndrome (SIRS) criteria.
"The problem with SIRS is it's good at identifying septic patients, but it's also good at identifying everybody who has a bad cold or a minor infection," Deutschman said.
Researchers knew that organs start to become dysfunctional in patients with sepsis and that the body's response is disproportionate. But that definition was too vague to help an emergency-room doctor making a diagnosis.
So two years ago the task force enlisted the help of Christopher W. Seymour, an assistant professor of emergency medicine at the University of Pittsburgh. Analyzing records from six million infected patients, Seymour identified those who died, or who survived after spending three or more days in intensive care.
"Then he started looking at what we would be able to measure early on that would identify those patients as rapidly as possible," Deutschman said. "What Chris did was a stroke of genius."
Seymour took every variable that anybody had ever looked at when considering sepsis and the early symptoms of sepsis.
"You can get a terrific predictive value with something as simple as an elevated respiratory rate greater than 22, a systolic blood pressure, or top number, of less than 100 [normal is 120], and a change in mental status," Deutschman said. "If you have two out of three of those, you have a very high likelihood of ending up in the ICU or dying with suspected infection."
To identify septic shock, task force member Manu Shankar-Hari from University College London looked at the sickest patients in the group and found that they had elevated serum lactate and needed drugs to bring up their low blood pressure.
"The combination of those two things identified a cohort of patients with mortality in excess of 50 percent," Deutschman said.
Since its publication, the study has been downloaded from JAMA's website more than 180,000 times and has had in excess of 875,000 hits. Most of the feedback has been positive, but the study has stirred controversy among those reluctant to switch from the old criteria.
"There are a lot of people who have concerns about the requirement for both blood pressure and lactate for septic shock," he said, adding that the task force's work needed more study. "That said, I think we now have better tools."