Date: February 1, 2013
There was a time, not so long ago, when the field of patient safety and quality didn’t exist. The tacit assumption was that when it came to safety, any field with the mantra of “Do no harm” wasn’t actually likely to do much harm. Was it?
Turns out, patients were dying by the stadium full. That’s what the Institute of Medicine announced in its groundbreaking 1999 report To Err Is Human: Building a Safer Health System. The report found that anywhere between 44,000 and 98,000 people died from medical errors each year in U.S. hospitals. The report thrust the issue of patient safety into the national spotlight.
For Hopkins Medicine, the report was a call to arms, brought home all too vividly just two years later. The death of young Josie King from a cascade of medical errors and miscommunications at Hopkins in 2001 put patient safety on the front page; Ed Miller, former medical school dean, said that “Josie’s death had a huge impact on the institution, and in some ways it allowed for a change of culture to occur.”
Across the institution, numerous clinicians and nurses began their own investigations into improving patient safety; like molecules in solution they needed only the right structure to crystallize and strengthen their purpose.
That would come through the Armstrong Institute for Patient Safety and Quality, launched in 2011, and its catalyst Peter Pronovost, winner of a MacArthur “genius grant.” An anesthesiologist and critical care medicine physician who had lost his own father to a medical error, Pronovost’s early work in safety included inventing an infection reduction checklist for catheter use in intensive care units. The checklist is today credited with saving thousands of lives annually.
Since then, Pronovost has worked to bring serious data and rigorous science to the field, building, with his colleagues, a growing archive of scholarship that is well on its way to achieving critical mass. To date, the Armstrong Institute’s more than 100 core faculty from the schools of Medicine, Public Health, Nursing, and Engineering, and Hopkins’ Applied Physics Lab have published more than 400 papers; the institute garnered $12 million in grants and contracts for the fiscal year ending in July. Perhaps most importantly, the Armstrong Institute provided safety and quality training to more than 650 patient safety advocates (including a half-dozen Patient Safety Fellows).
Drawing from fields ranging from computer science to psychology, the patient safety and quality effort at the Armstrong Institute is making an impact across Hopkins Medicine—and on hospital systems around the country.
“Our work is informed by human factors; systems engineers, behavioral economists, sociologists, health informatics … we learned when you package all those things together, you get a good conceptual model of how to change behavior and improve outcomes,” says Pronovost.
Here are five examples of how that’s playing out.
Preventing Diagnostic Error
Just as mariners once depended upon the art of reading the stars to set a true course, so, too, do physicians rely on a tool that is equal parts art and science to determine their course of action with a patient. It is called a diagnosis, and like those chart headings of yore, it’s often fallible.
In fact, outside studies corroborated by Hopkins researchers estimate that perhaps 10 percent or more of all diagnoses made by physicians are erroneous, often resulting in a spectrum of problems that range from the delayed start of useful treatments to increased morbidity and mortality. A team of researchers including Hopkins’ Bradford Winters and David Newman-Toker estimated that diagnostic errors contribute to the deaths of more than 40,500 people annually in the ICU—roughly equal to the yearly mortality toll from breast cancer.
A classic example of a common diagnostic mistake: When a patient presents with breathing difficulties in the ED, is it pneumonia or acute respiratory disease syndrome? One calls for medications, the other a ventilator. The wrong call, either way, leads to poorer outcomes even if it’s course-corrected fairly quickly. Take the diagnosis a step further: If one diagnoses pneumonia, which kind? Bacterial, which calls for antibiotics, or the less common fungal, which requires different meds? Again, mistakes are disturbingly the norm, says Winters. “What surprised us was how often a certain kind of fungal pneumonia, aspergillosis, was discovered at autopsy that was not diagnosed [nor treated] while the patient was in the ICU; it was an eye-opener how common that was.”
Newman-Toker, who studies how dizziness as an early warning sign for stroke is often missed, notes several cognitive factors involving physicians that can lead to diagnostic errors. There’s “availability bias,” which is the effect of one’s most recent clinical experience. “If you’ve seen something recently, you’re more likely to rank it as likely to occur again. If you get burned by a rare disease, you’re going to overorder tests for the next three years to ensure you don’t get burned again,” he says. “It happens in overestimating uncommon things, and underestimating the likelihood of common, dangerous diagnoses. We constantly don’t do the math right.”
One way to overcome personal bias, especially in ICU settings where patient status is constantly changing, is what one Massachusetts-based study called “diagnostic huddles.”
“They had criteria that if a patient wasn’t progressing, a huddle could be called,” says Winters. “Bring in the wisdom of the crowd and figure out if they were in a cognitive trap and not valuing certain data to either raise things up or lower things down on the differential diagnosis scale.” Huddles have been credited with improved scheduling, reduced testing, and better preparation for addressing patients’ specific needs.
Computers may eventually help decide which data is valuable to huddlers. The sophisticated algorithms that now allow national security agencies to sift millions of data points for patterns pointing to usable intelligence could be brought to bear on patient charts, says Winters. Have enough records of, say, 65-year-old men presenting with breathing difficulties, and a slight spike—or perhaps drop—in temperature might be enough to send an alert of what the true diagnosis should be.
There may be other tip-offs as well. Newman-Toker’s team is looking at properly assessing stroke in the ED, especially in patients who present initially with severe dizziness. It turns out that measuring these patient’s eye movements is more accurate at predicting a stroke than even MRI scans of the brain. “It could be transformative for these stroke patients,” he says, “ and someday, we might even be able to assess the eye movements just using the camera portion of an iPhone.”
Help for “Second Victims”
Albert Wu never forgot his first medical error, or the support he received in its aftermath. As a medical student, he accidentally gave a patient an overdose of morphine. The patient immediately went into respiratory arrest but was resuscitated and turned out to be fine. Wu wasn’t. He wondered if this would be his last patient, until the nurse manager on the unit pulled him aside. She assured him he was a solid young doctor with a promising future, shared with him a mistake she had made in her career, and gently pointed out how he could prevent such mistakes in the future. “I was treated very generously by her; she said, ‘Don’t be so hard on yourself.’ It completely changed my reaction to the interpretation of incident,” says Wu.
Indeed, as Wu, an internist at Hopkins, began research on medical errors (he was an early advocate for full disclosure of errors to patients, and sat on the board that wrote Hopkins disclosure policy in the late 1990s), he never forgot that medical errors affect staff as well as patients. He even coined a term in the literature—“Second Victims”—for staff traumatized by being witness to or participants in a medical error.
Most concerning to Wu was the culture of medicine that called for maintaining a façade of emotional calm regardless of whatever one has seen or done. It simply wasn’t human—or humane. Staff needed an outlet for their pain; Wu and his team have given them one. It’s called RISE—the Resilience in Stressful Events team. It’s a completely anonymous, volunteer peer support effort.
RISE was rolled out in November 2011 in the pediatrics department. “Most of the calls involved the death of a patient, but not necessarily an adverse event,” says RISE coordinator, nurse Cheryl Connors. “For people who experience these events, it’s so painful, and they don’t know what to do. The fact that they can call someone, a peer who is listening and reassuring them that they’re good people, provides them comfort and encouragement.”
Modeled in part after a University of Missouri program that has fielded more than 600 calls, Hopkins’ RISE program is just starting out, with a planned expansion throughout the hospital in 2013. Whether the program, which isn’t counseling so much as what Wu calls “psychological first aid,” offers a tangible benefit to callers has yet to be measured. But there’s little doubt that, without an outlet, staff can carry emotional scars for years.
Connors, who was part of the team caring for Josie King when King died from a preventable medical error in 2001, noted that “the standard response on the unit was not to talk about any of it; not her name, the situation, any of it.”
The silence had a price: Within a few years of the event, many of the nursing staff had left the unit. Connors invited a few of the nurses to be part of the RISE program and they talked about the pain they kept locked inside. “One nurse said she thought about that little girl every day of her life for a good six years. She has a niece with the same name. Every time she sees her, she can’t get Josie King out of her head for a month.”
Wu says research shows that disaster relief workers given peer counseling appear to benefit from the support; he’s hoping the RISE team will offer similar assistance, and keep staff from burning out over unresolved emotional issues surrounding patient events. The initiative also has implications for patient safety. With no outlet for their pain, Wu says, “the performance of these clinicians can be impaired as a result, potentially leading to further safety problems.”
DrPH candidate Hanan Edrees, who helped create and administer the staff surveys that showed a need for what became RISE, is encouraged by staff response. “There’s a lot of research saying clinicians working on units have the same [stress] burden as people in a war zone. They don’t realize they are carrying that much stress. What we, their peers, can tell them [through RISE] is that they are human, and the pain is normal.”
Reducing Alarm Fatigue
There’s no arguing that staff on most ICUs face an “alarming” situation: A Hopkins study published in 2010 found that on the 15-bed medical progressive care unit, the staff faced nearly 1,000 alarms per day, with anywhere from 80 to 99 percent being of a nonactionable nature. This nonstop cacophony of, essentially, false alarms leads to a psychological state known as “alarm fatigue,” where staff begin to ignore alarms completely.
Reducing the number of unnecessary alarms and lessening alarm fatigue was the charge of the Alarm Management Task Force at Hopkins, which began work in 2006. Through a combination of engineering changes and a greater level of nursing autonomy, the task force was able to drop alarms on the medical progressive care unit by 43 percent, as reported in 2010 in the American Journal of Critical Care. Notably, the changes, such as engineering pagers to receive alarms, resulted in no apparent additional risk (and perhaps a benefit) to patient care.
Study author Maria Cvach says a major change was allowing nurses to manually alter the point at which certain alarms go off, depending upon the condition of a patient. As an example, changing oxygen saturation alarms to go off at 88 percent saturation instead of 90 percent greatly reduced alarms, with no harm to patients. Similar reductions involved changes in the parameters for high and low heart rate alarms. Cvach has shared the data at several national conferences, helping to spread the word about the potential benefits of alarm reduction.
The task force was so successful that it recently received the annual health devices achievement award from the ECRI Institute, a patient safety research nonprofit. And the work is now standardized and implemented in all Hopkins medical units: “We prompt nurses to set parameters at the beginning of their shift,” says Cvach. “It’s right on the Electronic Patient Record: ‘Have you checked your alarms?’”
ICU of the Future
The basic footprint of an intensive care unit room hasn’t changed much since its ancestor, the Iron Lung wards, first made their appearance during the polio scourge of the 1930s and ’40s. Over time, caring for the sickest of the sick evolved, but everyone was too busy to give much thought to how the parts related (or didn’t) to the sum total of care, says Hopkins intensivist Bradford Winters.
“We never [added technology] under the guise of what we wanted the room to really look like and function,” he notes. “It was more that as new technology came in, we just stuck it on the wall.”
That thinking is now evolving, as researchers work to figure out how clinicians, families, and caregivers can best interface with emerging technology to create a safer, more effective ICU. A $9.4 million grant from the Moore Foundation will allow Hopkins physicians and outside partners to imagine the “ICU of the Future” from the ground up. Everything from the physical layout of rooms to how people interact with equipment—and each other—will be in play, and the effort will involve experts from engineering, infection control, systems analysis, psychology, safety and quality control, and more.
One of the more fascinating players in the joint effort is Lockheed Martin, which is bringing its experience in defense and aerospace technologies to the clinical medicine environment. Lockheed’s expertise in getting multiple systems to, in essence, “speak” to each other in moments of crisis—as when a plane is trying to evade an incoming missile—may provide exactly the kind of inter-device communication, data analysis, and pre-crisis rapid response currently absent in today’s ICUs.
Simulation and training tools already in use in aerospace are “two or three generations ahead of what the health care industry is using,” says Lockheed’s corporate director of healthcare initiatives, Robert Szczerba, who sits on the Armstrong Institute’s board.
He imagines ICU training that would provide a fully immersive, virtual experience: “Using virtual environments [such as headgear], you create the experience of actually being in that ICU room. With avatars that can be controlled by other individuals [they can even be off-site], you could have events that trigger alarms, or have a patient’s electronic patient record create a virtual event that you, as a clinician, could respond to.” Such immersive training, he suggests, would be more effective “than the way it is now in hospitals, where you have nurses learning infusion pumps from thick manuals instead of a simple smartphone app.”
While Szczerba and Co. are looking at engineering and computer solutions, Adam Sapirstein, a critical care specialist at Hopkins, is working with human factors engineers and bioethicists at the Berman Institute to examine the social and psychological factors that humans bring to bear on the ICU setting.
Sapirstein is focused on the interplay between a trio of “burdens” faced by ICU staff and how families and patients might play a role in easing all three: the “cognitive burden,” the sheer ability to cope with the inherent stress of the ICU environment; the “procedural burden,” in terms of skills and being able to step into the swirl of activity and triage what steps to take in what order; and the “burden of disease” itself.
Sapirstein uses the example of a kidney transplant patient: While the operation is nearly routine, the follow-up ICU care is anything but. “Sometimes they have no venous access sites left because of long-term dialysis. And they may have diabetes and hypertension, and we can’t give them medicines they need through a venous catheter… there are many ways to have mistakes made in their care.”
For his part, Sapirstein’s colleague Dale Needham believes that the ICU of the future would be “like an inpatient rehab unit for critically ill patients where they could get comprehensive rehabilitation therapy all while on life support.” The critical care specialist has done studies showing that minimizing sedation and getting patients physically active in the ICU even while on a ventilator is the best way to avoid delirium and maintain vital physical function.
“There’s clear evidence,” says Needham, “that if patients get this approach to care they’re more likely to go home from the ICU rather than to a rehab facility.”
Putting Family on the Menu of Care
Nurse Rhonda Wyskiel still vividly and painfully recalls the time 20 years ago when she wasn’t allowed to care for her mother in the ICU. Wyskiel was in nursing school when her mother was hospitalized. She had been used to taking care of her mom at home, but suddenly that caregiving was verboten and Wyskiel was furious.
“I was told I couldn’t wake her, or touch her. I couldn’t wipe her face with a washcloth or brush her hair. Yes, the nurses were in there, pushing all the right buttons, but they weren’t doing the things that were important to my mom. My mom loved having chapstick on her lips. No one ever did that. I asked if I could do it. They said they would, but they never did. I didn’t know I could speak up. My mom ended up passing away and I didn’t get to spend the kind of time with her I wanted to. I was angry about it, and thought, ‘I can’t let this happen to anybody else.’”
Informally, she did just that when she became a nurse, working on units that embraced family involvement as a part of the patient and family centered care movement. But she wanted to do more, to send a clear visual signal that family help in caregiving was not only welcomed but actively encouraged for a host of tasks. Her solution takes the form of the Family Involvement Menu, a large, green, dry-erase board in each patient room of the Weinberg ICU at Hopkins.
The board is brilliant in its simplicity. For families, it’s a guide to numerous activities they can assist in, such as helping to turn a patient in bed or washing a loved one’s hair. For nurses, it’s a helping hand, one they definitely need.
“One of the barriers we have [to providing complete care] is that we don’t have enough time to implement all the things we’re asked to do in a day,” says Weinberg clinical nurse specialist Samantha Young, who assisted Wyskiel in the development and roll-out of the menu. “I always tell nurses to utilize their resources, and the families are a great resource. As Rhonda discovered when she polled families on the unit, everyone wanted to do something, they just didn’t know how they could help.”
Now they do, in ways that are more vital to outcomes than they might imagine. Young notes that early mobility following surgery is often a key to getting out of the ICU quickly, but nurses don’t always have time to walk their patients.
“One thing on the menu is ‘range of motion’ for increasing early mobility,” says Young, who has spoken about the family involvement initiative across the country, including before the Michigan Hospital Association. “When I lecture, I have a slide of one of Rhonda’s families, walking like a herd around a patient, in the morning, with coffee cups in hand. The people I lecture to have an overwhelming favorable response because they’ve never seen anything like the menu and they’d love to implement it.”
The menu also helps with infection control, by encouraging family members to provide oral care for the ventilated patient (which involves teeth brushing with an antiseptic or hydrogen peroxide)—important for preventing ventilator-associated pneumonias from developing. Oral care should be provided six times a day, but a nurse on a busy shift may miss a treatment.
In the Weinberg ICU, the menu looks like it’s here to stay. The protocol process is being built into electronic patient charts, which prompts nurses to both engage families in the menu and document caregiving activities in which families take part.
There is also research being planned to determine the menu’s effectiveness in improving measurable outcomes, such as reducing pain and decreasing ICU delirium. “I’m working hard on the evaluation piece and also building a readiness assessment into it,” says Wyskiel, who cautions that the units best able to take on the menu are those that have “a strong family-centered culture already.” *
Fatal Flaws in the ICU
- Each year, 40,500 patients in U.S. intensive care units die with a missed diagnosis that may have contributed to their death.
- 28 percent of all adult ICU patients have at least one misdiagnosis at death.
- In 8 percent of patients, the error was serious enough to cause or directly contribute to death.
- Compared to adult hospital patients overall, ICU patients face up to a twofold
risk of suffering a potentially fatal diagnostic error.
SOURCE: “Diagnostic errors in the intensive care unit: a systematic review of autopsy studies,” BMJ Quality and Safety; Bradford Winters, lead author.