First of two parts
Fresh from surgery, the patient was wheeled into the intensive care unit and immediately hooked up to a cardiac monitor that would alert nurses to a crisis. Sometime during the following days, though, the cables running from her chest to the machine slipped loose.
The monitor repeatedly sounded an alarm — a low-pitched beep. But on that January night two years ago, the nurses at St. Elizabeth’s Medical Center in Brighton didn’t hear the alarm, they later said. They didn’t discover the patient had stopped breathing until it was too late.
At Tobey Hospital in Wareham, nurses failed to heed a different type of warning on a September morning in 2008. An elderly man’s electrocardiogram displayed a “flat line’’ for more than two hours because the battery in his heart monitor had died. While nurses checked on him, no one changed the battery. The man suffered a heart attack and was found unresponsive and without a pulse.
These were just two of more than 200 hospital patients nation wide whose deaths between January 2005 and June 2010 were linked to problems with alarms on patient monitors that track heart function, breathing, and other vital signs, according to an investigation by The Boston Globe. As in these two instances, the problem typically wasn’t a broken device. In many cases it was because medical personnel didn’t react with urgency or didn’t notice the alarm.
They call it “alarm fatigue.’’ Monitors help save lives, by alerting doctors and nurses that a patient is — or soon could be — in trouble. But with the use of monitors rising, their beeps can become so relentless, and false alarms so numerous, that nurses become desensitized — sometimes leaving patients to die without anyone rushing to their bedside. On a 15-bed unit at Johns Hopkins Hospital in Baltimore, staff documented an average of 942 alarms per day — about 1 critical alarm every 90 seconds.
In some cases, busy nurses have not heard or ignored alarms warning of failing batteries or other problems not considered life-threatening. But even the highest-level crisis alarms, which are typically faster and higher-pitched, can go unheeded. At one undisclosed US hospital last year, manufacturer Philips Healthcare, based in Andover, found that one of its cardiac monitors blared at least 19 dangerous-arrhythmia alarms over nearly two hours but that staff, for unexplained reasons, temporarily silenced them at the central nursing station without “providing therapy warranted for this patient.’’ The patient died, according to Philips’s report to federal officials.
In other instances, staff have misprogrammed complicated monitors or forgotten to turn them on.
The Globe enlisted the ECRI Institute, a nonprofit health care research and consulting organization based in Pennsylvania, to help it analyze the Food and Drug Administration’s database of adverse events involving medical devices. The institute listed monitor alarms as the number-one health technology hazard for 2009. Its review found 216 deaths nationwide from 2005 to the middle of 2010 in which problems with monitor alarms occurred.
But ECRI, based on its work with hospitals, believes that the health care industry underreports these cases and that the number of deaths is far higher. It found 13 more cases in its own database, which it compiles from incident investigations on behalf of hospital clients and from its own voluntary reporting system.
Massachusetts does not have a searchable public database of device problems. But the Globe was able to identify at least 15 deaths in the past six years that appear related to missed alarms or problems with monitors that track heart rate and rhythm, blood pressure, blood oxygen levels, and other physiological functions, based on reports, many redacted, provided by the Department of Public Health and the Centers for Medicare and Medicaid Services. Thirteen of those cases involved cardiac monitors, and at least seven instances involved dying batteries or disconnected cables.
“Yes, this is real, and, yes, it’s getting worse,’’ said Carol Conley, chief nursing officer for Southcoast Health System, which includes Tobey Hospital. “We want to keep our patients safe and take advantage of all the technology. The unintended consequence is that we have a very over-stimulated environment.’’
‘Alarms all the time’ In the St. Elizabeth’s death, state investigators cited “alarm fatigue’’ as a factor and said staff “did not have a sense of urgency to address low- and mid-level alarms,’’ according to the report on the case. At Tobey, investigators found that the nurse caring for the patient who died had a “challenging’’ workload and did not “clearly ask for help when she needed it.’’
The most widely publicized case was in January 2010 at Massachusetts General Hospital, when an elderly man suffered a fatal heart attack while the crisis alarm on his cardiac monitor was turned off and staff did not respond to numerous lower-level alarms warning of a low heart rate.
Ten nurses on duty that morning could not recall hearing the beeps, which sounded at the central nurses’ station, or seeing scrolling messages on three hallway signs that would have warned them as the patient’s heart rate fell over the course of 20 minutes and finally stopped.
State investigators said “nursing staff interviewed on the patient care unit said they were experiencing alarm fatigue and a desensitization to alarms after hearing them constantly throughout the workday.’’ Investigators did not say why the crisis alarm was off.
“That case sent a major ripple through the industry,’’ said George Mills, a senior engineer for the Joint Commission, an organization that accredits hospitals. The case became widely known after the Globe reported on it last year.
Mass. General would not allow the Globe to talk to nurses involved in the case, but nurses at other hospitals said the number of alarms can be so overwhelming that they turn into background noise — although a dozen nurses interviewed by the Globe said they have never seen a nurse purposely ignore an alarm.
“Everyone who walks in the door gets a monitor,’’ said Lisa Sawtelle, a nurse at Boston Medical Center. “We have 17 [types of] alarms that can go off at any time. They all have different pitches and different sounds. You hear alarms all the time. It becomes . . . background.’’
The different pitches and frequencies of the alarms are meant to signal the severity of the problem, with the most-piercing, fastest beeps generally reserved for potentially fatal heart arrhythmias. Generally, alarms go off both in a patient’s room and at a central nurses’ station, though wireless heart monitors worn by the patient sound only at the nurses’ station. Some hospitals, especially those that have had problems with missed alarms, have put speakers in hallways to make it easier for nurses to hear alarms. Heart rhythm, heart rate, and other parameters also are displayed on screens, usually in the room and at the station — and sometimes on brightly colored scrolling signs in corridors.
Hospitals that have experienced alarm-related deaths have aggressively improved monitoring.
Mass. General and Southcoast hired nurses and technicians whose sole job is to monitor the monitors. Southcoast also employed ECRI to evaluate its cardiac monitoring program and recommend improvements, while Mass. General is trying to reduce false alarms by making some monitors less sensitive to unimportant changes in patient health and tightening standards for when to put patients on monitors. St. Elizabeth’s replaced its ICU monitors and required bioengineers to check daily that they are working properly.
But the response overall from regulators, manufacturers, and many hospitals has been uneven and in some cases slow.
Manufacturers, which are required by the FDA to investigate these deaths when hospitals report them, almost always conclude that their monitors worked fine and that hospital staff misprogrammed them or didn’t respond to their beeps. But according to health care engineers and hospital staff, the devices do have flaws that can contribute to alarm fatigue.
Monitors can be so sensitive that alarms go off when patients sit up, turn over, or cough. Some studies have found more than 85 percent of alarms are false, meaning that the patient is not in any danger. Over time this can make nurses less and less likely to respond urgently to the sound.
“If you have a system that tells you nine times out of 10 there is no reason for that alarm,’’ said Dr. Michael Imhoff, a managing director of Boston MedTech Advisors, a Dedham consulting firm, “it totally desensitizes you.’’
‘My uncle deserved better’ By the time a very ill Edward Harrigan was rushed to Tobey Hospital in September, 2008, he felt his life’s mission was done. Harrigan’s wife, Virginia, had suffered from advanced Alzheimer’s disease, and because they had no children, he was her lone caregiver. Despite her condition, he took her out for breakfast daily, eventually buying a recreational vehicle to make those short drives more comfortable.
Virginia Harrigan died Sept. 2, and in the following days her husband stopped eating and taking care of himself, said his nephew, David Harrigan.
Once at Tobey, Harrigan, 87, wore a wireless heart monitor, called a telemetry unit. According to state investigators, the battery inside his monitor began to weaken at 7:21 a.m. on Sept. 22 and died completely at 9:31 a.m. — causing a flat line to display on a monitor at the nurses’ station. “The patient’s EKG displayed a flat line for 2 hours and 10 minutes and no one noticed and/or addressed it,’’ the state found, although nurses said they continued checking on him.
The malfunction might not have mattered, except that Harrigan’s heart shut down. Because the monitor was not working, no alarm sounded and no one was with him when he died.
“My uncle deserved better and so does every patient at any hospital,’’ David Harrigan said. “Even if my uncle was older, fragile, and distressed, no one should be left unattended or cared for with faulty equipment. That is just wrong.’’
Southcoast has taken aggressive steps to improve its monitoring system, including hiring monitor-watchers — at a cost of $1 million the first year — implementing standard volumes for all alarms, and changing batteries at a set time every day. When Harrigan died, the system did not sound an audible alarm for dying batteries. It does now.
The Globe, with the help of ECRI, reviewed copies of 216 reports that device manufacturers filed with the FDA on deaths in which problems with monitor alarms occurred. The names of hospitals and patients had been removed for privacy reasons.
Initially, hospitals almost always blame the monitor’s alarm for not sounding when it should have, according to the reports. But the company investigations show that assertion is often false.
In 49 of the cases where hospitals said there was no warning, when companies reviewed the monitors’ internal record of alarm activity, they found the monitors did work. Often the volume was too low to be heard — sometimes intentionally kept low by nurses to protect themselves and patients from excessive noise. At one undisclosed hospital in 2009, General Electric found that a cardiac monitor in an intensive care unit was set at 40 percent of full volume, with the result that no one responded to an arrhythmia alarm for 40 minutes before the patient died.
In other instances, it is unclear why staff did not hear or respond to alarms.
In 25 cases, companies found that when alarms sounded, staff either accidentally or intentionally silenced them — sometimes without providing care to the patient that could have prevented his or her death. Alarms are sometimes silenced for legitimate reasons — the nurse has checked on the patient and he or she is fine, or the patient has indicated that he or she does not want medical intervention in the event of heart failure. But sometimes they are silenced because a nurse assumes the alarm is false, said nurses and hospital safety specialists.
In 40 of the cases reviewed by the Globe, the alarms did not sound, usually because staff had not properly programmed or turned on the machines; in only eight cases was there a malfunction or design flaw. For example, after one death, manufacturer Masimo Corp., based in California, decided to recall a pulse oximeter that provided only a “visual alarm’’ for a defective sensor; its new units have an audible alarm that should be more noticeable to staff.
But in nearly 100 cases, manufacturers were unable to determine exactly what went wrong, often because they didn’t have enough information, or they told federal regulators they still were reviewing the death.
Hospital staff said that in eight of the 216 cases the monitor problem did not result in delayed care.
Experts say companies don’t always get back to the FDA with the results of their case reviews — and the agency doesn’t always follow up — making it nearly impossible for the FDA to determine the extent of alarm problems.
“These reports are not thorough,’’ said James Keller, vice president of health technology evaluation and safety at ECRI. “It’s very hard to have a good idea of exactly what’s going on.’’
In Pennsylvania, the state with the most robust reporting system for medical errors, a separate analysis by ECRI for the Globe found 35 deaths related to physiological monitor alarms since the state began collecting data in June 2004; 28 of these cases involved a “staff failure,’’ including nine instances that involved alarm fatigue.
Cynthia Lacker, a patient safety analyst with ECRI who oversees the Pennsylvania data, said that even if staff had heard and responded to these alarms, it’s unclear whether the patients would have survived with speedy medical intervention; many were elderly and extremely ill.
“Some of these deaths are probably preventable, but we don’t know for sure,’’ she said.
Looking for solutions In February 2002, the national organization that inspects and accredits hospitals issued an urgent alert. A review of 23 patients who died or sank into a coma because their ventilators malfunctioned found that in 65 percent of the cases, staff didn’t respond to alarms warning that machines were malfunctioning, set alarms incorrectly, or didn’t hear alarms set at low volume.
The Joint Commission, the hospital accreditor, cautioned hospitals not to become too dependent on alarms and advised them to routinely test alarms and better train staff on how to use them. In 2004, the organization decided to make these recommendations part of its national patient safety goals, which signaled they were a high priority.
But the very next year, the commission dropped the alarm advice from the safety goals, believing the problem had been largely solved, based on visits to hospitals.
Yet, the problem with alarms of all kinds hadn’t gone away at all. As the Joint Commission moved the issue to the back burner, hospitals and companies were continuing to report alarm-related deaths to the FDA. The number of hospital patients on monitors climbed as the monitors’ use expanded from intensive care units to regular floors.
Last winter, Mills, of the Joint Commission, got an earful from a gathering of biomedical engineers from hospitals across the country. He said they told him that in recent years, hospital staff have disconnected monitor speakers, taped over them, and turned down volume, all to escape the constant noise.
Joint Commission executives said they now realize their initial efforts were inadequate given the complexity of the problem.
Like the Joint Commission, the FDA has occasionally tried to educate providers about the alarm problem. The agency has posted at least two “device safety’’ bulletins on its website in the past decade, warning providers not to ignore alarms and not to forget to turn them on, and last month distributed a video on alarm safety to 4,500 hospitals and nursing homes.
But ECRI’s Lacker said education is the “weakest intervention,’’ and nurses say something must be done to reduce the number of false alarms.
One paper published in 2006 in the American Journal of Emergency Medicine found 99.4 percent of alarms in one hospital emergency room were false, and for patients with chest pain, fewer than 1 percent of alarms resulted in a change in care for the patient.
Dr. Roger Mark, head of the Laboratory for Computational Physiology at MIT, found 43 percent of crisis alarms, which warn that the patient is in immediate danger, were false in a review of medical records for 447 patients in the intensive care unit at Beth Israel Deaconess Medical Center between 2001 and 2007.
These patients were not in danger at all, but had coughed, turned in bed, or experienced a tremor, or some other “artifact’’ had temporarily altered their heart rate or rhythm. In other cases, the signal from the cables attached to their chest was simply poor.
“Some days you feel you are just running from alarm to alarm. It can be exasperating,’’ said Sylvia LaRocca, a nurse for 37 years at Beth Israel Deaconess.
Hospital staff complain that false alarms also lead to higher costs and anxiety for patients. In November 2007, one undisclosed hospital complained to the FDA in its Medical Product Safety Network — a voluntary program intended to uncover medical device problems — about a GE cardiac monitor.
The monitors, which also measure blood oxygen levels, “alarm at the slightest modification,’’ suggesting a baby is sicker than he or she really is, the hospital reported, causing “a possible longer length of stay, worry for parents, and additional tests.’’ The company would not comment, but the FDA said GE changed the software to allow hospitals to delay the alarm 5 to 15 seconds, giving the machine more time to determine whether the baby really is in trouble.
Manufacturers say that they are improving their machines, but that they make them sensitive because neither they nor caregivers want to miss a real crisis.
“Missing a real event is much more costly to the manufacturer,’’ Mark said, referring to the potential for a lawsuit. “On the other hand, if a nurse fails to look at a monitor when a beep goes off and the patient dies, that’s the nurse’s fault.’’
The FDA, meanwhile, has not forced industrywide changes in design to cut down on alarm fatigue, because the solutions are unclear, said Dr. Brian Lewis, a cardiologist and medical officer in the FDA’s division of cardiovascular devices.
“We are at the point where we all know [alarm fatigue] is a problem,’’ he said. “We know we’d like these devices to serve the public health better. But we need specific direction from providers on how these devices should be changed.’’
Solutions that seem obvious, he said, could create new problems. For example, after the Mass. General patient died last year, many safety experts suggested that manufacturers should make it impossible for critical alarms to be turned off by hospital staff.
Philips is leading an industry working group on alarms that is trying to agree on standards — and the group rejected this approach because of caregiver objections, said Mike Mancuso, vice president and general manager for patient monitoring systems.
“Some audio alarm signals must be capable of being made quiet for a period of time to permit the clinical staff to solve the problem in peace and quiet,’’ he said.
Mancuso argued that the technology “is getting better but that this problem is not going to be solved by technology alone. It is the timely and appropriate response to the alarm that is the issue.’’
Liz Kowalczyk can be reached at [email protected]