The first thing you notice are the alarms.
Walk into the intensive care unit in just about any American hospital, and you’ll be bombarded with beeping and blaring noises and flashing lights. It may look high tech. It’s not.
It’s “no different than it was 50 years ago,” said Dr. Peter Pronovost, a critical care physician at Johns Hopkins Hospital in Baltimore. “There are stacks and stacks of machines with wires sticking out of them. It’s chaos.”
ICUs are one of the most crucial departments of any hospital — heroic places with devoted staff who pull the sickest of patients from death. But many ICU physicians say they’re also woefully — and often dangerously — out of date. Six million patients in the United States pass through ICUs each year, and studies show serious and sometimes fatal medical errors are routine. And a recent review published in the journal Critical Care found no major advances in ICU care since the field’s inception in the 1960s.
Now, a handful of doctors and nurses in places like Baltimore, Boston, and San Francisco are trying to yank the ICU into the 21st century.
Pronovost, for instance, has called in submarine engineers and the physicists who built the spacecraft that whizzed past Pluto last year to help redesign intensive care. They’ve been shocked at how primitive even new ICU units can be: “They walk through the ICU and just flinch,” Pronovost said.
One of the most pressing problems: None of the medical devices so critical to patient care — ventilators, pumps, drug infusers, pulse rate monitors — talk to each other, and, in what’s dubbed “the alarms race,” all try to outdo each other by beeping ever louder. Nurses answer a false alarm on average every 90 seconds, he said.
“We have alarm fatigue. We’ve become numb to the noise and start to block them out,” said Rhonda Wyskiel, a former ICU nurse at Johns Hopkins Hospital who now works to develop patient safety measures for hospitals at the Armstrong Institute for Patient Safety and Quality, which Pronovost directs.
Pronovost, who made his name devising a checklist for doctors to consult before inserting a central venous catheter — a simple innovation that dramatically cut bloodstream infections — is now trying to create a “smart ICU.”
“What I want to do for the ICU is what Steve Jobs did for the iPhone,” he said.
Moving beyond paper protractors
To take one example: Many patients’ beds need to be raised to prevent pneumonia. Nurses are supposed to check angles each shift, sometimes using paper protractors. It’s a vital check that sometimes isn’t done or is not documented. Pronovost’s solution: a $2 sensor that could monitor the angle of the bed continuously.
Another sensor could monitor the compression devices that are supposed to pump patient’s limbs to prevent deadly blood clots but are often left unplugged. He’d also like to connect ventilators to patient medical records to make sure vital information like the patient’s height — which affects the ventilator setting — is transferred. In his ideal world, all devices in the ICU would be networked and continually monitored, cutting the cacophony of alarms and the nursing workload.
ICU nurses face an average of 200 duties per shift and spend a lot of time checking and double-checking orders and logging simple data from one device into another. Devices that actually spoke to each other and integrated information would leave them more time to spend with patients instead of machines, Wyskiel said.
Such thinking is long overdue, said Dr. Marie Csete, an anesthesiologist and critical care specialist who now heads the Huntington Medical Research Institutes in Pasadena, Calif., and has coauthored a series of papers proposing ICU upgrades in the Journal of Critical Care. She said patient monitoring “needs to be designed from the ground up” because the current outmoded system is dangerous, inefficient, and impersonal.
“Somewhere under the ventilator, rapid infuser, and stacks of drips, there is a patient, but where do you put your gaze?” said Csete. “We’re creating a generation of doctors who look at screens instead of patients.”
Csete knows the dangers first hand: A few years ago, her mother, who was 87 at the time, nearly died in a Florida hospital after an aortic valve replacement. While in the ICU, she developed multiple organ failure and pneumonia. Orders were ignored, alarms were disregarded, and no one seemed to be paying attention. “If I hadn’t been there, she would have been dead the second day.” Csete said. “I was not impressed.”
But reengineering an ICU is a huge undertaking, and one that involves a skill set not taught in medical school. “When physicians see the amount of math involved, they just scatter to the hills,” Csete said.
The paralyzing weight of data overload
Take the problem of data overload.
All those noisy devices in the ICU generate an immense amount of data. In a modern ICU, a single patient can generate 2,000 data points per day, said Dr. Brian Pickering, an anesthesiologist and critical care physician at the Mayo Clinic in Rochester, Minn. In a 24-bed ICU like his, that’s 50,000 data points a day. Important information is easily lost, or forgotten.
Pickering joined the Mayo Clinic nine years ago from Ireland, where patient data was still logged on a paper chart at the end of the bed. He was overwhelmed, he said, by electronic records in the United States that had too many tabs and screens and were difficult to navigate.
“Point. Click. Point. Click. Point. Click. Back and forth,” he said. “That may work if you’ve only got one patient. But I’ve got 24 in the ICU, and any one of them could be in crisis at any minute.”
With colleagues, Pickering created an “electronic intern,” called AWARE, that identifies the most important information a physician needs and highlights it, organizing it around organ systems. (The system is now being sold to hospitals through a Rochester startup called Ambient Clinical Analytics; Pickering and the Mayo Clinic benefit financially from the sales.)
Another app now being tested, called EMERGE, extracts data from patient records to warn clinicians if an intervention they are planning might cause harm.
“There’s so many things physicians can’t find, so things get missed,” said Hildy Schell-Chaple, an ICU nurse at the University of California, San Francisco Medical Center who has been testing EMERGE, which was developed at Johns Hopkins.
Yet another approach comes from Brigham and Women’s Hospital in Boston, which is testing a new secure microblogging platform that allows everyone on an ICU care team to see all messages relating to a patient. It promotes better communication between staff and the patient, and ideally, leads to fewer errors.
That is, if people use it, said Dr. Anuj Dalal, the hospitalist who designed the program. He said some staff members think of the system as too much work. They prefer email, or even old-fashioned pagers.
“From a technology standpoint, it’s usable,” Dalal said. “Getting people to use it is a completely different thing.”
Learning to see the patient as a person
Similar resistance has slowed the adoption of telemedicine, which can link specialists trained in critical care medicine to small hospitals lacking such expertise. The remote specialists can order treatments, check prescriptions, detect errors, and even talk directly to patients.
Early on, some physicians and nurses on the ground in ICUs so disliked the feeling of being watched by distant experts that they threw lab coats or towels over cameras. Slowly, acceptance is growing; telemedicine systems are in place in about 16 percent of the country’s ICU units, said Dr. Craig M. Lilly, a critical care specialist at the University of Massachusetts Medical School and expert on telemedicine.
A 2014 study showed telemedicine can reduce ICU mortality — in large part by ensuring that nurses and physicians respond quickly when patients take a turn for the worse. “Sometimes the nurse doesn’t notice, sometimes the nurse is busy doing other things, sometimes the nurse is too chicken to wake up the doctor at 2 a.m., and sometimes the doctor just won’t listen to a nurse,” Lilly said.
New technology may also help ICUs — once notorious for alienating families and keeping them at arm’s length — better include loved ones in a patient’s care.
UCSF is now testing bedside tablets that patients or families can use to upload photos and descriptions of themselves. They can let doctors know what they like to be called, what their hobbies are, what they fear about their hospital stay, and what their healing goals are. The care team can then see them as individuals — and not, Schell-Chaple said, as just some 48-year-old man in Bed 8 who had a liver transplant.
“The ICU environment,” she said, “is not set up to treat people with respect and dignity.”
‘A eureka moment’
The biggest hurdle to building a truly smart ICU, has been medical manufacturers who don’t want to open up their devices and share the data they collect.
“I used to be guilty of that too, and it’s unfortunately so shortsighted,” said Joe Kiani, founder of Masimo (MASI), a manufacturer of noninvasive patient monitoring devices based in Irvine, Calif. “We all think we have this amazing data and we want to hoard it, thinking we’ll monetize it some day.”
But Kiani soon came to realize that free data flow and linked devices were key to improving patient safety. He founded the Patient Safety Movement Foundation in 2013 and is working to get medical device manufacturers to sign pledges that they’ll share data from their devices.
Some 60 of about 100 key device manufacturers have signed on, said Kiani, who is an electrical engineer by training. He’s spent the past decade working on a device, called Root, that can collect and simplify multiple streams of patient data.
“I now see a future where everything’s connected,” he said. “Hospitals are finally having a eureka moment.”
Correction: An earlier version of this story misstated the location of Ambient Clinical Analytics.