Big Data and Alarm Management - Marshaling a Brain Trust to Find Answers

Big Data and Alarm Management - Marshaling a Brain Trust to Find Answers

Imagine the chaos for firefighters in a firehouse if the alarm rang incessantly. What is worse, imagine if only a small percentage of the alarms were real emergencies. The firefighters would be exhausted after only a day or two and their mental preparedness would surely suffer.

After reacting to the onslaught of false alarms, the reaction of the firefighters would eventually be like dealing with the proverbial boy who cried wolf. The danger posed by this situation is that when a legitimate emergency comes along; it can be a life-or-death situation. This holds true in the hospital setting.

The hospital is no Aesop’s Fable; it’s a place where constant awareness is demanded. Clinical alarm safety deserves the attention of all the stakeholders because the most vulnerable stakeholder is the patient.

A cacophony of alarms can hide the real emergency.

It is likely that none of this is news to anyone reading this article. The issue is a frontrunner among issues that will be addressed by experts across the health care spectrum this year and in the future.

On ECRI Institute’s “Top 10 Health Technology Hazards for 2015,” the top spot was “Alarm Hazards: Inadequate Alarm Configuration Policies and Practices.” In 2013, The Joint Commission (TJC) issued Sentinel Event Alert Issue 50, “Medical Device Alarm Safety in Hospitals.”

AAMI’s National Coalition for Alarm Management Safety held its initial meeting in April of 2014. Like cybersecurity, EMRs and dealing with newer regulations from CMS, the topic of alarm management requires new thinking and the adoption of intelligent guidelines.

Although a majority of alarms will never require clinical engagement, the risk to patient safety requires that attention be paid to all of them. Many will be sounded with no basis for an urgent response. There are some recurring reasons for these non-significant alarms.

As TJC and AAMI have pointed out, many alarms result when “conditions are set too tight; default settings are not adjusted for the individual patient or for the patient population; ECG electrodes have dried out; or sensors are mispositioned.”

With devices monitoring a patient’s blood pressure, heart rhythm, blood oxygen level or ventilator status, the variables become a consideration. Normal blood pressure for a person in their 80s may be different than that for someone in their 30s, for instance.

Alarms on cardiac monitors, pulse oximeters, ventilators, CPAP/BiPAP devices, infusion pumps, feeding pumps, bed exit alarms, chair exit alarms, vacuum assisted closure devices, hypo/hyperthermia machines, SCD machines, intra-aortic balloon pumps and continuous renal replacement therapy devices all add to the orchestra of confusion.

The Joint Commission’s National Patient Safety Goal (NSPG) on alarm management recognized that the story from the fable is rooted in a real problem. No real alarm can ever be missed. At the same time, the multitude of false alarms tire out clinicians to the point of mental exhaustion; alarm fatigue.

The old adage; “no man is an island,” has application in this instance. No hospital should be an island when an issue effects thousands of hospitals. The issue has created a paradigm that finds patient safety at its core everywhere.

Mining for Tweaks

The challenge of managing alarms becomes somewhat more vexing when you compound it with the goal of integrating big data into a solution. The move from paper to electronic records was a monumental task in itself and many health care providers aren’t there yet. Data mining this growing universe of information, and assigning it an additional, and new utilitarian purpose,

is the next chapter in the health care environment evolution. If just one nurse can be spared one headache, it is a step in the right direction.

The monumental amount of data collected these days has helped to provide some key insights into what can be done to reduce the number of alarms that don’t require immediate attention.

Looking forward, there are changes that must be incorporated into the protocols of hospitals everywhere.

“Major changes are needed to improve alarm management. Vendors need to improve alarm specificity — the trueness of alarms,” says Maria Cvach DNP, RN, FAAN, assistant director of nursing, clinical standards at Johns Hopkins Hospital and chair of AAMI’s Clinical Alarms Steering Committee.

“This may mean better alarm algorithms that manufacturers use to test their devices; and incorporating short delays prior to an alarm sounding to allow for alarm auto-correction when a HR or Pulse Ox momentarily changes due to artifact or patient movement,” Cvach says.

“We need multi-parameter device analysis to decide when an alarm should sound versus crossing a threshold momentarily. This means that we need for devices to use more than a single threshold breach to determine when to sound,” she says.

Like many health care systems, Rex Healthcare in Raleigh, North Carolina, took action to address the alarm management challenge.

“We began with collecting and reviewing our equipment database with select individuals from various areas of clinical responsibility; the committee then determined which systems and equipment should be targeted for improvement and which clinical areas having these systems would benefit most from our efforts to enhance the overall clinical alarm environment,” says Steven Bowers, CET, manager of Biomedical Engineering at Rex.

“The obvious choices for us became the critical care areas of Cardiac Intensive Care Unit, Medical Intensive Care Unit and Cardiac Thoracic Intensive Care Unit. We conducted alarm assessments and collected actual alarm data previous to imposing any intervention to validate our decision to focus on these clinical sites and departments,” he says.

Once they were validated, the biomedical engineering team ordered and acquired updated software for its monitoring assets, installed, tested, and then worked closely with clinical staff and vendors to drill down for the best alarm protocols for each area.

“Admittedly we went back a few times to make adjustments to protocols and monitoring set ups before we felt we had achieved the best case scenario for patient care while presenting a more effective and efficient alarm environment,” Bowers adds.

He says that soon after the changes were made, the department began sampling and taking data off its network to provide a clear picture of what the initiative had achieved.

“While everyone working in the ICUs had noticed a considerable change to their working environment, the data reports concluded that we had exceeded expectations. Our clinical ICU staff all report a much more staff- and patient-care-friendly environment and one that delivers a more safe, effective alarm notification system overall,” Bowers continues.

Indeed, during a 14-day review period, both pre- and post-intervention and after software upgrades, alarms had dropped anywhere from 26 to 48 percent in the units where the problem was confronted.

Click here to read this entire article on TechNation!