Baseline assessment was key to reducing alarms
In 2006, Maria Cvach, MSN, RN, CCRN, assistant director of nursing clinical standards at The Johns Hopkins Hospital and Andrew Currie, MS, CBET, the director of clinical engineering, were asked to head a team to reduce clinical alarms. But first they had to define the problem. Just how many alarms were going off, and what kind?
They put together a small task force that began analyzing alarm systems on one unit using a quality improvement rapid cycle change approach.
“Our first challenge was learning how to analyze alarm system data,” Cvach says. “It took us two years to figure out how to extract the right data. It’s hard to believe it took that long, but that’s how difficult it was.”
Currie was able to create a real-time surveillance system to integrate data feeds at the bedside from multiple medical devices. “We were suddenly seeing unfiltered data from our GE monitors, and we saw an unbelievable number of alarms,” Currie says.
By observing the alarm condition patterns, they identified that many of the conditions were clearly false. For example, apnea alarms were coming from patients on ventilators. Experience with this system drew their attention to the high volume and the inaccuracy of alarm conditions coming from their monitors.
Adam Sapirstein, MD, associate professor in the Department of Anesthesiology/Critical Care Medicine in The Johns Hopkins School of Medicine and also a faculty member in the Johns Hopkins Armstrong Institute for Patient Safety and Quality, was involved in the pilot project. “It could have been one full-time person’s sole job to just silence all of those alarms,” he says. To him, the patient safety implications of so many false alarm conditions were clear.
At the same time, Cvach and the task force were focusing on the number of alarm conditions the nurses were encountering in their unit, which she says were “astronomical.” She worked with Currie to access the data, and a valuable partnership was formed. “You need quantitative data to evaluate the applications of alarm management in hospitals,” says Currie. “Our initial efforts to generate data were very basic.”
The GE monitors they used had a pager system that ran off of a local area network, he says, and a server listened to messages from the monitors and kept a log when it sent messages to pagers.
“I began looking at that log and tracking the changes that our alarm group instituted to measure the success of our efforts,” Currie says.