24 May 2011—According to doctors at the University of Massachusetts Medical Center in Worcester, intensive care units backed up by off-site doctors and nurses, who could remotely monitor critically ill patients and direct the ICU’s on-site staff, had fewer patient deaths and shorter ICU stays. Their trial of a so-called tele-ICU system, which allows intensive care specialists outside the hospital to see and hear patients, monitor vital signs, and access medical records, proves that such a system actually benefits patients.
Over the two-and-a-half-year study, off-site doctors and nurses manned multimonitor computer stations from a nearby building, where they received real-time information on patients. The UMass tele-ICU system is based on Philips’ Visicu eICU technology. The system’s software can detect trends that lead to patient deterioration. Off-site teams could verify these trends and, using microphones and cameras in each ICU room, collaborate with bedside nurses and physicians to treat the underlying causes.
The study’s results associate the use of tele-ICUs with lower mortality rates, shorter hospital and ICU stays, and lower rates of preventable complications. According to the findings, which were published last week in the Journal of the American Medical Association, the ICU mortality rate was 10.7 percent before tele-ICUs were introduced, compared with 8.6 percent afterward, and the mean length of an ICU stay was 6.4 days before tele-ICUs were used and 4.5 days after they were introduced. Before the tele-ICU experiment began, 13 percent of patients developed ventilator-associated pneumonia, but during the tele-ICU trial, only 1.6 percent experienced that preventable complication.
Craig Lilly, who led the UMass team, says that the success at UMass would be hard to translate to all ICUs. The academic medical center setting was a good backdrop for coordinating an effective program, and executive support also helped to integrate the technology into the staff workflow, he says. That was important, because the system "isn’t really about the computer but rather the interaction between the physical technology and the people at the bedside." The complex eICU technology is ultimately intended to enable better, more-efficient teamwork, he says.
With previous tele-ICU research, the technology’s benefits had been unclear. For example, a 2009 study from the University of Texas at Houston was conducted across multiple hospitals, with relatively low levels of staff cooperation. "The differences in the results are likely due to the fact that all of the patients in [the UMass] ICUs were allowed to have the tele-ICU physicians care for them at any time; in our study, only a third did," says Eric Thomas, the lead researcher for the Texas study and a past collaborator of Lilly’s.
Brian Rosenfeld, chief medical officer for Philips and a cofounder of Visicu, says that the ICU staff involved in the Texas study didn’t optimize the eICU technology, which he thinks was a factor in its more ambiguous results. He says that the UMass team, on the other hand, "demonstrated how to use the tele-ICU system effectively." The correct operation of the system depends a lot on how committed doctors are to using the technology and the authority they grant remote doctors to intervene in patient care, Rosenfeld says.
Jeremy Kahn, a professor who works on critical care issues at the University of Pittsburgh, has similar thoughts about successful tele-ICU use. He says the UMass study "provides the first convincing evidence that ICU telemedicine can be an effective complement to bedside care in some settings." In a commentary he wrote to accompany the publication of the UMass study, he emphasizes that UMass had positive results because the use of a tele-ICU system was a part of wider quality-of-care improvements that the medical center was making.
It is difficult to isolate the unique benefits of the tele-ICU from a hospital’s ongoing quality-of-care improvements. Kahn argues that although we now have more evidence that telemedicine can be used effectively in the ICU, the results don’t demonstrate that specific improvements in critical care—such as following best practices or reducing preventable complications—were directly linked to the tele-ICU system. "This is important, because many of these practices could just as easily be improved through intensivists (critical care specialists) at the bedside," he says.
Thomas agrees: "The medical intensive care unit in my hospital has gone three years without a ventilator-associated pneumonia. Rates of other complications are also very low," he says.
However, Rosenfeld counters, there’s a shortage of critical care specialists in the United States, so adding bedside staff would be difficult. Instead, he thinks the eICU technology can make more efficient use of existing staff.
But an eICU system doesn’t come cheap. The start-up cost for 100–120 beds, across multiple hospitals in a health-care network, is about US $5 million, Rosenfeld says. That’s too much for many institutions, but UMass recouped their investment before the end of the first year using the system, largely due to the shorter ICU stays. This ultimately led to a higher patient volume, because each ICU bed is in high demand, Lilly says.
Lilly and his colleagues acknowledge that separating the effects of an ICU’s general quality-improvement measures from the direct benefits of remotely monitoring patients was a challenge, but some of the changes they witnessed during the trial were easy to interpret. For example, family members of patients quickly understood that the tele-ICU unit could continually monitor a loved one’s status, which made them much more comfortable going home periodically to rest, Lilly says. Another clear result was improvement of care at night, which is usually considered to be of lower quality.
Ongoing upgrades to the eICU system might help distinguish its impact from traditional bedside care, Rosenfeld says. In particular, Philips is developing more meaningful graphical data displays, more efficient displays of data from multiple patients, and more intelligent processes for interpreting the data.