Snitkin’s skill is using computer programming to make sense of biological data. He thought sequencing the KPC could provide the kind of “fingerprint” Palmore needed to determine if a single strain was circulating.
A month earlier, Snitkin, Segre and Palmore had met for the first time to discuss collaborating on a study about hospital-acquired infections. Now that they knew KPC was spreading, the topic was considerably more pressing.
Information in DNA can be understood as a code of chemical bases that appear in different combinations of four letters—A, G, C and T. Humans have 3 billion of these bases, while a bacteria like KPC has 6 million. Scientists read the code by having a computer analyze the chemicals, write out the letters and then compare them to one another. This is called sequencing the genome. It took $3 billion and 10 years to sequence the human genome in 2001. Now it takes just a few thousand dollars and a few weeks.
Reading code comes naturally to Segre, who grew up in Philadelphia in a family of physicists and mathematicians. Her grandfather worked in military intelligence during World War II and broke Nazi codes, and her father is a particle physicist. One of Segre’s earliest memories is of her grandfather creating coded messages as a game to entertain her and her siblings.
“He would do things like write messages about where to find chocolate in hieroglyphics or symbols, and then we’d write back to him in code about things like, ‘Let’s go for a bike ride,’ ” she says. “I used to think all grown-ups could read codes.”
Segre went on to study mathematics at Amherst College in Massachusetts, graduating in 1987 and then earning her Ph.D. in genetics at the Massachusetts Institute of Technology. There, she worked with Eric Lander, one of the leaders of the Human Genome Project, the government-funded effort to map the human genome. She joined NIH in 2000 and is married to a conservation biologist.
Since Segre had already been thinking about MRSA and hospital-acquired infections, she readily agreed to Snitkin’s plan. On Aug. 24, Segre emailed Palmore and suggested sequencing KPC.
“I didn’t even know if [the sequencing] could be done or if it would be useful,” Segre says, “but it was: Let’s try it and see. What would be the harm?”
At an NIH outpost in Rockville, a whole floor is filled with computers and machines that sequence genomes for NIH researchers. After several weeks of work to determine the protocols for sequencing KPC’s DNA, Segre sent the bacteria of Patients One through Four for analysis.
On Sept. 22, the team learned that all of the KPC strains were related. Segre was “stunned” by the result and its implications.
For the next three months, Palmore and Henderson worked with the NIH staff to stem the outbreak. Each time a new patient tested positive for KPC, Palmore would call Segre and Snitkin to sequence the genome. Each time, the strain came back related to Patient One, and the two epidemiologists would deploy another infection control strategy.
The leading way disease spreads in a hospital is via the hands of health care workers. But washing hands and wearing gloves properly before touching a patient can be a challenge in a busy and distracting health care setting. The compliance rate on washing hands in hospitals can range from 40 percent to 58 percent, according to the Hand Hygiene Resource Center in New Haven, Conn.
“There are a lot of things to do, and it’s hard to do them all correctly 100 percent of the time,” says Henderson, who has worked at the NIH hospital since 1979.
Palmore says the hand-washing rate at the NIH hospital was “high” in the summer of 2011, but declines to give a number. Her team started requiring staff entering and leaving patients’ rooms to wash their hands twice. They also hired “hall monitors” to ensure staff members complied. The hospital swabbed the hands of staff members and the environment around patients to see if the KPC bug was living on bed rails or sinks. Nothing turned up.
Palmore and Henderson implemented an even more drastic tactic that they’d decided upon at the end of August. They built a separate intensive care unit and staffed it with its own set of physicians, nurses and cleaning staff. Workers in those rooms weren’t allowed to touch their pagers or their cellphones until leaving. Eventually that isolated ICU would have its own X-ray and ultrasound machines. Patients who tested positive for KPC but didn’t have symptoms were also separated with their own staff. The NIH estimates that it spent at least $400,000 in staff time on these measures.
“This was very hard on staff because it required a lot of extra hours,” Palmore recalls.
Palmore and Henderson also sought outside guidance. Henderson spoke to colleagues at the CDC and other hospitals. The hospital expanded its testing to a rectal swab, which is uncomfortable but more accurate.
That move proved important: It would catch four more cases in September.
An environmental swab that month also was key. It showed one patient’s ventilator was contaminated with KPC even after sterilization. The bug was hardier than anyone had anticipated.
The ventilator, which had only been used on Patient Six, spurred the infection team to take another step. The hospital hired Bioquell, a Horsham, Pa., company that decontaminates hospital rooms and equipment with robots that spray a hydrogen peroxide vapor.
Still, the bug was spreading.
By the end of October, 12 patients had tested positive for KPC. Five had developed a bloodstream infection and would die.
“Since no one could pinpoint what, in each instance, had transmitted the bacteria or what equipment, it was hard to know who to be angry at,” Palmore says. “We would start to feel optimistic when no case had been identified—and then we’d be crestfallen” with each new case.
She and Henderson could only speculate on how the KPC was spreading. A nurse or a visitor could have lost his or her balance when leaving a patient’s room, they posited, and inadvertently contaminated a wall. Or an environmental swab could have failed to detect it on a wall or a cart and someone could have accidentally transferred it to another patient. “For any of these transmissions, it’s probably small, unrecognized failures with consequences,” Palmore says.
Meanwhile, Palmore heard from a colleague that hospitals had found KPC surviving in the drains, so that November she ordered sinks to be tested. Six, all in rooms with KPC patients, were found to be contaminated. The hospital ripped out the sinks and plumbing and replaced them.
“We just needed to keep at it,” Palmore says of all these strategies. “This is what happens with hospital transmission. You can’t just turn [the transmission] off, because a lot of it has already happened.”
Beginning in mid-November, every patient in the hospital was tested for KPC to ferret out silent carriers. The tests picked up four more cases outside the ICU. Each patient was isolated.
“There was no stone left unturned by the hospital,” says Arjun Srinivasan, the CDC’s associate director for health care-associated infection prevention programs, who consulted with NIH on the outbreak.
That December, NIH officials contacted the Maryland Department of Health to discuss the outbreak. But the state failed to notify Montgomery County officials—and because of that, the hospital would come under heavy criticism when it made the outbreak public in August 2012.