Walk into a hospital intensive-care unit and hear the din: A ventilator honks loudly. An infusion pump emits a high-pitched beep-beep every six seconds. A blood pressure monitor pushes out one long tone after another.

This particular racket, at MedStar Washington Hospital Center, comes from medical devices and equipment that scans for potentially dangerous ­changes in patients’ heart rhythm, blood pressure and other vital signs.

But most of the noises are false alarms or don’t require action. The ventilator sounds a warning because a patient coughs. The infusion pump beeps after running out of a medication the patient no longer needs. The blood pressure monitor goes off after a nurse adjusts a catheter in the patient’s artery.

VIDEO: How an arterial catheter works

The sheer number -- several hundred alarms per patient per day — can cause alarm fatigue. Nurses and other workers, overwhelmed or desensitized by the constant barrage, sometimes respond by turning down the volume on the devices, shutting them off or simply ignoring them — actions that can have serious, potentially fatal, consequences.

Clinicians and patient-safety advocates have warned of alarm fatigue for years, but the issue is taking on greater urgency as hospitals invest in more-complex, often-noisy devices meant to save lives. Last month,the Joint Commission, which accredits hospitals, directed facilities to make alarm safety a top priority or risk losing their accreditation. The commission is requiring hospitals, starting in January, to identify the alarms that pose the biggest safety risks by unnecessarily adding noise or being ignored. By 2016, hospitals must decide who has the authority to turn off alarms.

“We are saying, ‘Here’s an issue that hurts and kills people,’ ” said Ron Wyatt, medical director of the commission’s health-care-improvement division.

A muted monitor

The parents of Mariah Edwards won a $6 million malpractice settlement after their 17-year-old daughter died last year following a tonsillectomy at a Pennsylvania surgery center. After the surgery, the high school junior was given a potent painkiller that slowed her breathing. By the time nurses checked on her 25 minutes later, she had suffered profound and irreversible brain injury. She died 15 days later.

A nurse said in her deposition that the alarm on the respiratory monitor was muted, said Joel Feller, an attorney for the family. After Edwards’s death, the center announced several ­changes, including that alarms would no longer be muted.

The Joint Commission received 98 reports of alarm-related incidents — including 80 deaths — in the 31/2-year period ending in June 2012. In more than 60 percent of the cases, alarms were either inappropriately turned off or were not audible in all areas. Those voluntary reports are a gross undercount, says the commission, which estimates that there were close to 1,000 alarm incidents in which patients died or were injured, or faced those risks.

The ECRI Institute, a Pennsylvania-based patient-safety organization, listed alarm hazards as the No. 1 issue on its annual list of the top 10 health-technology dangers for 2012 and 2013.

“I think the main reason is the large growth in the use of monitors that have alarm-based features and the number of alarms that clinicians are needing to deal with,” said Jim Keller, ECRI’s vice president for health-technology evaluation and safety.

At Children’s National Medical Center, a late response to a heart monitor played a role in the death of a patient 10 years ago. That prompted changes, according to Linda Talley, vice president of nursing for critical care, who along with other officials declined repeated requests for details about the patient’s death.

A special team at Children’s Hospital determined that clinicians heard an alarm every 66 seconds, on average, in the hospital’s neonatal ICU. The team now scrutinizes monthly data to look at the number of alarms and the response times. To reduce non critical alerts, staff members customize device settings for individual patients. A heart monitor for a newborn, for example, would probably be set to alarm more often than that for an 18-year-old.

Because each manufacturer’s device alarms are different, the hospital also standardizes equipment as much as possible; one vendor supplies all the heart monitors, for instance.

Technicians in non critical-care areas keep track of flashing lights and other visual alarms on a central monitor and text the urgent alerts to nurses’ mobile phones.

Warnings on ‘complacency’

Patient-safety experts and health officials have known of the potential harm for years.

At Johns Hopkins Hospital, “it took a few adverse patient events to focus attention on the problem,” according to a medical technology group’s report about Hopkins’s safety efforts. The incidents took place in 2006, but officials declined repeatedly to say how many there were and whether patients died.

And Maryland health officials warned all hospitals in the state in August 2006 about “alarm complacency” after receiving reports of eight cases involving alarms — including four deaths — in the previous two years. None of the hospitals were named.

“It is sobering that critical care units, to which a gravely ill patient is sent for intensive monitoring, may fail to provide a safe environment,” the health department note said.

In one case, an elderly woman’s heart-rate monitor beeped unnoticed by staff for several minutes. Her nurse was preparing medication in another room where she could not hear or see the patient or the monitoring system. (The hospital had cut the position of the technician whose job was to monitor alarms at a central station.) By the time the nurse responded, the patient had suffered irreversible damage, and she died shortly afterward.

In another case, cardiac electrodes became disconnected from a patient who fell getting out of bed, triggering a specific alarm that usually indicates that the wires are not connected. Medical workers discounted it because it was considered a low priority. No one responded until housekeeping staff discovered the patient on the floor 30 minutes later. Clinicians were unable to resuscitate the patient.

Hopkins created an alarms task force, analyzed data and found that the average number of alarms that sounded per bed per day in one ICU was 771.

“That was just astounding when we saw those numbers,” said Maria Cvach, Hopkins’s assistant director of nursing, who co-chairs the task force. About 80 percent of the alarms were for low-priority conditions, she said. Yet “those are the ones that ring continuously and people tend to just ignore.”

Working with each unit’s medical directors, the task force turned off the sound for low-priority alarms. It also instructed nurses to adjust device settings for all patients during every shift. Heart-monitor electrodes that are placed on patients’ chests were also changed daily to keep them from drying out. Noise and false alarms fell dramatically.

The quieter environment also built better relations with patients and their families, who lost trust in staff members when they didn’t respond quickly to a one-beep alarm. Most patients don’t know that one-beep alarms are typically not as urgent as those with three tones, Cvach said.

“But sustaining this stuff over time — it’s a battle,” said Andrew Currie, Hopkins’s director of clinical engineering and another leader of the alarm task force.

Experts say there should be better reporting of “near misses” or incidents that could be dangerous.

The FDA steps in

The Food and Drug Administration received 862 death reports associated with alarms in the seven years that ended Dec. 31, 2012, an agency spokeswoman said. But the reports provide little information about what happened and don’t necessarily mean that alarms contributed to the deaths. They include cases in which a patient was dying and an alarm sounded.

“It’s often very hard to get to the bottom of what is going on,” said Mary Logan, president of the Association for the Advancement of Medical Instrumentation, a nonprofit organization focused on developing medical technology.

Two years ago, the AAMI hosted an alarm summit; the FDA, the Joint Commission, the ECRI Institute and the American College of Clinical Engineering took part. A summit report identified the need for smarter technology that can check more than one indicator and weed out nuisance alarms. Manufacturers, however, are reluctant to invest in more expensive technology because of liability.

“If there’s a mistake,” Currie said, “who’s to blame?”

The FDA, which regulates medical devices, is increasing staff awareness of alarm safety when it reviews applications for new devices. The agency is also working with industry, hospitals and other groups to standardize alarm sounds, especially for life-critical events, said William Maisel, deputy director of science and chief scientist at the FDA’s Center for Devices and Radiological Health.

And the agency is “actively preparing” a recommendation on ventilators, he said, that would for the first time ask manufacturers to address alarm reliability, including false alerts.

‘It can be very, very loud’

On a recent weekday at MedStar Washington Hospital Center in Northwest Washington, all 14 beds in the ICU for brain injuries are full. Most patients are hooked up to several pieces of equipment. A heart-rate monitor beeps nonstop until a nurse adjusts the setting to reflect the patient’s current condition. Moments later, a different beep sounds. The electrode on a patient’s heart monitor has slipped off, explains Charlie Sederstrom, a nurse manager.

Most alarms are adjusted to sound at 70 percent of full volume, and nurses are supposed to check the settings for each patient during each shift.

Still, “it can be very, very loud,” said Sandra Walsh, the unit’s assistant nursing director. Alarms also sound when patients are calling for help in the bathroom and when there is an immediate emergency situation with a patient.

In the hospital’s non-ICU areas, officials have been implementing a “no-pass” policy begun last fall. No employees, from housekeepers to administrators, should pass by a patient’s room if they hear an alarm. They need to make sure the patient is breathing and call for help if necessary.

Sue Eckert, chief nursing officer, said she does not know of any alarm-related patient deaths in her more than 30 years at the hospital. But she knows that the potential exists every minute of the day.

“It’s an enormous issue,” she said. “We’re as at-risk as everybody else.”

Alice Crites contributed to this report.