Alert fatigue leads to OR fatalities

Alert fatigue can lead to behaviors in health care that might seem fine until the day they cause a tragedy, says John Banja, PhD, assistant director for health sciences and clinical ethics at Emory University in Atlanta.

Banja was involved in the investigation of a sentinel event involving a patient who died in surgery. The patient was undergoing an abdominal surgery, and the operating room team had turned off all the alarms on the equipment, because they found them annoying, mostly because they went off too often, at too low a threshold. They had turned them off before many times before with no negative consequences, Banja says.

The procedure was proceeding well, and at one point the physician needed to obtain an X-ray of the patient's abdomen.

Because the patient's breathing would blur the X-ray, the surgeon asked the anesthesiologist to turn off the respirator while the X-ray was taken — not an uncommon request. The anesthesiologist turned off the respirator and the X-ray was made.

The procedure continued, but the anesthesiologist never turned the respirator back on. Because the alarms on the respirator, and everything else, were turned off, no one realized the patient was suffocating until it was too late.

"The patient was left in a vegetative state and died 11 days later," Banja says. "Here was a catastrophe that happened because of the way that a defense mechanism had been disabled in this hospital. The anesthesiologist made a slam-dunk error, but if the alarms had been functioning, they would have caught that error immediately, and you wouldn't have this horror occurring."

Clinical alert fatigue remains a vexing problem for health care providers, and the risk to patient safety is high. When clinicians become so annoyed by alarms that they disable them or so accustomed to hearing them that they do not respond appropriately, patient's lives can be at stake.

Part of the problem is that modern medicine just has too many beeps, buzzers, bells, and warning signs flashing on the screen, says Linda Peitzman, MD, chief medical officer of Wolters Kluwer Health in Indianapolis. If too many alerts are triggered when medications or tests are being ordered, for example, the likelihood is high that physicians eventually will tune out or actively override even high-severity alerts, she says.

Physicians may bypass CPOE

A Wolters Kluwer Health white paper on clinical alert fatigue noted that alert fatigue also can cause physicians to bypass or remain skeptical of computerized physician order entry (CPOE), resulting in low adoption rates that impact outcomes and the hospital's return on its technology investment. "This is typically the end result when, as one hospital [chief medical information officer] noted, alert fatigue drives clinicians to view CPOE as a challenge to their autonomy or the flavor of the month," the report says.

Alert fatigue is particularly prevalent with medication orders. Medication alerts are so common that they have created a situation in which "systems and the computers that are supposed to make physicians' lives better are actually torturing them," the white paper says. A 2009 study of nearly 3,000 prescribers in three states found that physicians ignored alerts more than 90% of the time, a rate that varied little based on severity.1 That figure is consistent with earlier studies, such as one in 2003 that showed prescribers overrode 80% of the medication alerts triggered in a hospital practice and another in 2003 that found prescribers overrode 91.2% of drug allergy alerts and 89.4% of the high severity drug interaction alerts.2,3

Some alerts unnecessary

The research suggests that some alerts are unnecessary and only serve as a nuisance to clinicians, but the white paper notes that nuisance alerts also might lead a facility to turn off entire alert groups, including some that might be relevant. Alert fatigue ultimately can generate dissatisfaction with the CPOE system as a whole, the report warns.

ECRI, the non-profit medical technology research group in Plymouth Meeting, MA, reports that alarm issues are among the most frequent problems reported by clinicians. ECRI advises that staff members must understand the purpose and significance of alarms and they should know how to set alarm limits to appropriate, physiologically meaningful values.4 "We continue to learn of incidents in which staff unintentionally disable critical alarms by setting them far outside reasonable bounds," ECRI writes. "Low-saturation alarms on pulse oximetry monitors and low-minute-volume or high-peak-pressure alarms on ventilators are regular subjects of this sort of error.

Peitzman says, "In my experience, alert fatigue is directly related to the level of sophistication and thought in how they were designed and implemented. This is not just a matter of people getting tired of hearing them or seeing the alerts pop up and overriding them just to make their lives easier. A lot of times the alerts are unnecessary, and in some cases inaccurate, and that can be a big contributor to alert fatigue."

The more often alerts activate, and especially the more often they activate inappropriately, the more likely a physician or other caregiver will override them, Peitzman says. Ideally, the alerts should be tailored to the particular patient, which would cut down on the number of inappropriate alerts, she says.

When you purchase a system

Alert fatigue should be considered when purchasing systems, especially CPOE and other systems that can include many alerts, Peitzman says. Look for systems that not only have good content in terms of what triggers an alert, but also offer the ability to customize the system to different situations, she suggests. (For more advice on reducing alert fatigue, see below.)

"Interruptive alerts" are particularly annoying when they are unnecessary, Peitzman says. This type stops the act of, for example, ordering medication until the physician orders the blood test required by the medication. In most cases, the physician must order the test or document why it is not being ordered before the medication order can continue. Peitzman says such alerts can be necessary, but they should be kept to a minimum. When the interruptive alert is not based on sound reasoning or does not apply to this particular patient or this type of care, the alerts can be frustrating and create time delays.

Alert fatigue also can be a problem with telemetry monitoring, says Kim Bonzheim, MD, director of the Heart and Vascular Center and Beaumont Hospital in Royal Oak, MI. Either with decentralized monitoring at nurses' stations or a centralized monitoring system for an entire hospital, the same problems arise, she says. Those monitoring the system hear the same alarms and see the same alerts so often that they can become inured to them, Bonzheim says.

Beaumont has a central monitoring system that works well, Bonzheim says. However, when a monitor sees a clinical alarm that needs to be passed on to the nurse, the alarm used to be sent by alpha-numeric pager. "Our problem was that it was a one-way communication device," Bonzheim says. "If the nurse didn't call to say she got the page, our system was set to keep sending that page every three minutes. As you can imagine, we drove the nurses crazy, and pagers were left on desks."

That system led to a near-miss and a sentinel event, Bonzheim says, so the hospital had to address the problem. The hospital switched to a two-way communication system that allows them to talk directly to the telemetry monitor. The time for completing a communication loop between the telemetry monitor and the nurse was cut from an average of about eight minutes to 39 seconds.

Deviation becomes the norm

Alert fatigue is another example of the normalization of deviance, says John Banja, PhD, assistant director for health sciences and clinical ethics at Emory University in Atlanta. When clinicians become accustomed to hearing alarms and seeing alerts, and when it becomes common and acceptable to override them, a wrong behavior has been normalized in the culture, he says.

"You start doing the short cuts and deviating from the protocols and the rules and regulations, and pretty soon you start exposing the patient to growing levels of risk," he says. "You don't even realize you're doing it and, in fact, you think you're doing it for the good of the patient, or for your own good and whatever enhances your own good is, in turn, good for the patient."

Managers should encourage a culture in which deviation is not accepted and alert fatigue is openly acknowledged as a potential problem, Banja says. The issue should be addressed in grand rounds, incident reports, weekly meetings with staff, and any other situations in which the problem can be discussed, he says.

Banja notes that such deviations are common in all industries. The recent BP oil spill in the Gulf of Mexico was traced, in part, to a series of deviations from standard procedure, including managers overriding alarms because they didn't want to wake up the crew who were sleeping, Banja notes. The deviations can seem innocuous or even a wise decision, until something goes wrong.

"Maybe you make all these risky deviations, but nothing bad happens and that convinces you more and more that it's OK," Banja says. "When you look at these disasters, what you [find] is that the deviations went on for months and years and nothing bad ever happened until the rig blows up, or you do a wrong-side surgery, or you give the patient an overdose of medicine. Then people get very bent out of shape and we start blaming and punishing, when we should have been holding them more accountable all along the way."

References

  1. Health Data Management. Avoiding 'Alert Fatigue.' Health Data Manag. Oct. 1, 2009. Available at http://www.healthdatamanagement.com/issues.
  2. O'Reilly KB. Doctors override most e-Rx safety alerts. Amednews.com. March 9, 2009. Available at http://www.ama-assn.org/amednews.
  3. Isaac T, Weissman JS, Davis RB, et al. Overrides of medication alerts in ambulatory care. Arch Intern Med 2009;169:305-311.
  4. ECRI. "2010 Top 10 Technology Hazards." Health Devices 2009;38:11-16.

Souces

For more information, contact:

  • John Banja, PhD, Assistant Director for Health Sciences and Clinical Ethics, Emory University, Atlanta. Telephone: (404) 712-4804. E-mail: jbanja@emory.edu.
  • Kim Bonzheim, MD, Director of the Heart and Vascular Center and Beaumont Hospital in Royal Oak, MI. Telephone: (248) 898-5000.
  • Linda Peitzman, MD, Chief Medical Officer, Wolters Kluwer Health, Indianapolis, IN. Telephone: (317) 735-5300.

Take these steps to reduce alert fatigue

Involve physicians in the development and implementation of alert systems, rather than simply training them in the systems when you're ready to go live, says Linda Peitzman, MD, chief medical officer of Wolters Kluwer Health in Indianapolis.

"Get their input into how and when they want these things to fire, so that when they do fire, they won't be surprised," Peitzman says. "They will understand why they were included, and they will be much more apt to pay attention to them."

With technology growing so rapidly and health care providers wanting to improve safety by standardizing processes, there is a temptation to keep adding more alerts, she says. "There are many other tools you have for clinical decision support [that can be added] into the work flow," Peitzman says. "So, ask whether another alert is the right thing. Maybe we could do an order set that automatically has this in it instead of reminding them. Or when they document it, this action automatically happens instead of reminding them to do it."

When implementing computerized physician order entry (CPOE) or any system with built-in alerts, the threshold for alerts should be turned down, at least initially, to ensure there is buy-in and not a perception that there's interference with the clinical workflow, says Dan O'Keefe, MD, executive vice president for the Society for Maternal-Fetal Medicine, chief medical officer for PeriGen, and an OB/GYN and maternal-fetal medicine expert. O'Keefe offers this further advice:

  • Introduction of clinical alarms as a part of any system should be a deliberate and gradual process. Otherwise, a hospital risks alienation of staff, who already are busy and trying to get up to speed on new technology and new behavior.
  • Alerts absolutely must be clinically relevant or the doctors and other staff will certainly ignore them.
  • The frequency of alerts and the use of hard stops, which prevent doctors from taking any further action, are important. Overuse will only cause frustration and disillusionment with the system. Choose carefully when deciding which interruptive alerts are necessary.
  • Alerts should be evaluated by hospital staff on a quarterly basis to make sure they are clinically relevant. If they are not relevant, but simply a nuisance, they may be removed. Conversely, if the alerts are relevant and support the practice of evidence-based medicine, then the hospital must instruct the doctors to comply.

Peitzman agrees that health care providers should monitor the use of alerts within their systems, looking for chances to improve. "You should not only measure how many of them are being acted upon, but you also should talk to your physicians and find out about the effectiveness, the value, the appropriateness, and you can continue to modify them over time," she says.

Also, don't be afraid to disable alerts that came as part of a system package, she says. Ask yourself if this alert is something truly important to your patients or if it will just add to the cacophony of alerts. A warning about dietary restrictions or food interaction might be valid, for instance, but is it important enough to justify an alert screen for the physician? Pare down the alerts to the ones that really matter, and clinicians will pay more attention to them, Peitzman says.