The Big Picture
A Roundtable Discussion
Alarm Safety: A
Mary K. Logan, Moderator
When an 89-year-old man died in his bed at prompted us as an organization to try to wrap
Mary K. Logan,
Massachusetts General Hospital and the alarms our hands around the whole issue of alarm
JD, CAE is
that should have sounded didn’t, the incident set fatigue. We just completed a study funded by
AAMI. Email: off a different kind of alarm throughout the one of our alarm manufacturing vendors on the
mlogan@aami. hospital, in Boston, and in state regulatory issue of alarm fatigue.2 We were aiming to
org agencies. The tragic situation also provided a partner with this vendor to increase the
learning experience for Mass General and the specificity and positive predictive value of the
larger medical community.1 In that spirit, AAMI information the monitors give us.
Angela Andrew- recently gathered a group of experts to discuss a
Webb, RN is number of alarm safety questions. What kinds of Steve Wilcox But I see two other problems with
private duty alarms are necessary, and how many alarms are too alarm systems. Besides the false alarms, the
nursing case second problem is the lack of integration
many? How can technology help solve this
problem rather than compound it? between devices. All these independently
Christiana Care developed alarm systems don’t talk to each
Health System. Mary Logan What are some common problems other or integrate in any way. So even if the
Email: aandrew-webb@christianac- seen with alarm systems for medical devices? false alarm problem was eliminated, there
are.org would still be this cacophony because of that
Linda Talley I’m responsible for looking at how lack of integration. The third problem is that
nurses interact with technology. The literature the signals themselves are poorly designed so
Cartwright, RN is
program director tells us, and our own experience at Children’s they’re not natural sounds. They’re difficult to
of private duty National Medical Center indicates, that we’re learn and identify, and they go out of their way
nursing with dealing with anywhere from an 85% to 99% to be annoying. I think it’s an artifact of the way
Christiana Care false positive rate on alarms. We are inundated they’re designed.
with information and alarms, most of which are
meaningless to us. Tobey Clark IEC standards for alarm systems3,4
Nurses become desensitized to the huge aim to prescribe a way to standardize alarms
J. Tobey Clark number of alarms they’re confronted with in in terms of priority and parameters so people
is director of their daily work and, as a result, critical or can learn and understand alarm signals, and
clinically significant events can be missed. We better recognize the higher versus lower prior-
services with had a serious patient event many years ago that ity alarms. The American College of Clinical
the University of related to the timeliness of our response to a Engineering’s Healthcare Technology Founda-
Vermont Techni- monitor. In our post-event review, we learned tion published a white paper about clinical
cal Services that it wasn’t an equipment failure. Rather, it alarms and one of their recommendations was
Program. Email: tobey.clark@its.
was a human factors failure. That really for clinical alarm standards.5
8 Horizons Spring
Single user license only. Copying, networking, and distribution prohibited. 2011
The Big Picture
Wilcox Of course these standards have made Jim Welch A study published in Anesthesiology6
things a little more rational. But they haven’t focused on a general care setting, Dartmouth
addressed the problems of integration or Hitchcock Medical Center, where my company
false alarms. worked directly with the end users. Rather than
study how many alarms were occurring, we
Talley Our experience, and our discussions asked, “How many alarms are tolerable from
over a three-year period with a vendor we were a human factors standpoint to avoid alarm
working with, indicates that device manufactur- fatigue?” The nurses told us two to four alarms
ers are challenged as well by a consumer base per patient per day.
that has asked them, over the past 10 years, I would encourage the community to agree to
to throw it all at us. We ask them to give us a common set of metrics and terminologies.
everything they can with these “magic Everybody knows what alarm fatigue is, but it
machines” and we’ll deal at the front line with needs to be defined. They took a different
how we are going to discriminate between approach at Dartmouth by actually separating
the massive amounts of noise we’re the alarm annunciation, or the sound of the
confronted with. alarm, because alarm fatigue is mostly an audio
So I think they are very anxious to partner rather than a visual phenomena. They set
with us to figure out how to strike that balance. alarms to a lower threshold for surveillance, Roundtable Participants
The machines we put in the hands of clinicians that is, letting you know when a patient needs
present serious challenges. The level of Dr. John Hedley-Whyte, MD is
to be rescued versus letting you know when a
the David S. Sheridan Professor of
decision support is increasingly sophisticated patient crosses the threshold. That is a very Anaesthesia and Respiratory
and complex for any one clinician to manage. different approach to the traditional conditional Therapy at Harvard University.
In many of our intensive care units (ICUs), we settings found in ICU settings. By adding a Email: john_hedley-whyte@hms.
have additional personnel doing the decision delay to it, they were able to achieve four alarms harvard.edu
support, trying to counteract the massive per patient per day. And because of that, they
amounts of noise the technology produces. were able to actually improve patient safety MS, RN is vice
In our study we recorded tens of thousands methods as measured by escalation of care to president of
of alarms in a 30-day period, which translated the ICU and rapid response activations. nursing for
to approximately 900 per day. In one of our On the technical side, monitoring companies Critical Care,
critical care units, a total of 39,000 alarms were Heart Institute
have followed this paradigm of alarms being
recorded in a 30-day period which equaled 1300 activated by the crossing of a threshold. Those Systems at
alarms per day, or one alarm sounding every 66 thresholds often have not been defined using Children’s
seconds. In another critical care unit, we an evidence-based, rationalized approach to National Medical Center.
observed approximately 600 alarms per patient alert a nurse or clinician to go to the bedside. Email: email@example.com
per day. In a study out of Johns Hopkins,7 they found
Jim Welch is
In direct response to the sentinel event we that by lowering an Sp02 alarm from 90% to vice president
had here at Children’s National several years 88%, they were able to reduce the occurrence of of patient
ago, we decided every patient in every unit was alarms by more than 50%. A methodology to safety at
going to be on a monitor. So we became even rationalize what’s an appropriate alarm is Masimo, Inc.
more acutely aware then of the dilemma of and vice presi-
needed. The whole topic of alarm fatigue
dent of ACCE.
having very little science to drive how we really begs the question of why are we doing Email: jwelch@
monitor alarm limits. alarms? What’s the primary purpose of them, masimo.com
We felt bound, from an ethical perspective, to and how do we create decision systems or
do no harm and monitor patients to the fullest filters so that nurses can focus on clinically Stephen B.
degree possible. But we find ourselves now in Wilcox is a
this conundrum where we encounter an
abundance of information that is of very little Logan What are the biggest obstacles to solving a human
value. So we have duplicative measures on all of challenges related to alarms? factors
our units in terms of humans with an eye on consultancy.
patients, as well as trying to filter through what Email:
Wilcox One is a strong bias toward false
the alarms are telling us. positives instead of false negatives. If an
alarm signal fails to annunciate when there is
Single user license only. Copying, networking, and distribution prohibited. Horizons Spring 2011 9
The Big Picture
a legitimate situation, the liability is obvious. On the other hand, ond audio delay on it, essentially a filter, we found that we could
when the alarm signal annunciates when there’s not actually a eliminate more than 80% of the alarms. So now we’ve got the
problem, there is a logical liability because it’s undermining the alarms people care about, the ones where levels fall and stay
value of the alarm. However, I’ve never heard of anybody being below a threshold for a sustained amount of time, that would
sued for a false positive. So to avoid even the thought of a false cause a clinician to rationally say, “I’m worried about
negative, we’re just inundated with that patient.”
In part, machines are going off because they
Welch There is a lack of research in this area. I’ve scoured the are too sensitive. With the probes we use in the
literature for publications on alarm recurrences, origins, and home setting, if the client moves a bit too much
causes, and I’m only aware of a few published studies: the ones or even if their extremities are cold, we see a lot
I’ve already mentioned and another in the Journal of Emergency
of false alarms.
The JEM piece studied the occurrence of alarms and what Angela Andrew-Webb I can offer a home healthcare perspective
were clinically actionable events. It was predominantly focused on the alarm fatigue issue. In part, machines are going off
on electrocardiogram (ECG) alarms, which is, I think, the because they are too sensitive. With the probes we use in the
primary cause of most alarms, along with impedance respiration home setting, if the client moves a bit too much or even if their
rate. In the emergency room, they found that less than 2% of all extremities are cold, we see a lot of false alarms. If ventilators are
alarms require a physician to do something at bedside to reverse not set up properly, we can have water in the lines, which will
the condition. give us false alarms also.
So what we don’t answer in the literature is, what’s being The payer source is our major obstacle to solving these
monitored? What’s the alarm mean? And what’s the profile of problems. For instance, with ventilators, some of the home
that alarm? We conducted a study in 10 hospitals looking at Sp02 medical equipment (HME) companies are only compensated to
alarm occurrences using our technology. We found that if you give us two vent circuits a month. That means the circuits are
set your Sp02 alarm at 90%, you will have a lot of true alarms. But only changed every two weeks. They need to be changed every
those true alarms do not require a clinical intervention. By week to function properly.
lowering the alarm level from 90% to 88% and putting a 15-sec-
Join Us Today
ASPPS seeks to advance patient safety as a distinct and
vital healthcare discipline—offering the only individual
membership society of its kind for patient safety
professionals, risk management and quality leaders,
medical and nursing students, patient safety advocates,
and all those committed to safe patient care.
Learn about membership benefits by
Save the Date Fourth Annual Lucian Leape Institute Forum & Gala For details, visit
September 22, 2011 Guest Speaker: Atul Gawande, MD, MPH www.npsf.org
Boston Best-selling Author. Surgeon, Brigham & Women’s Hospital
Single user license only. Copying, networking, and distribution prohibited. Horizons Spring 2011 11
The Big Picture
Welch If we look at this from a systems John Hedley-Whyte These problems have been
perspective, think about links in a chain. If you around for a long time. Many of the topics we’re
don’t have the correct or optimized sensor, then addressing here were also addressed in a 1992
you don’t have a high-performance sensor, and publication I edited.10 The bottom line is that
you’re going to generate false data. False data everyone is to blame for problems with
leads to false alarms. If you’re trying to use the alarm systems.
same cable over and over and it’s at the end of The use of technology to allow intelligent
its life and its signal is becoming intermittent, escalation of alarms up the management chain
you’ll get false alarms. and alarm retrievability is essential. We need to
So from a provider medical community, how tap into the literature from other disciplines on
do you optimize every link in that signal chain? this topic, such as aeronautics.
Because the more you measure, the more The liability issue is a big one here. When
alarms you’re going to have. Once you have you talk to vendors, purchasers, trustees, and so
optimized the sensor, its placement, the skin forth, they all come back behind the defense of
prep, and you’ve got a good signal, now you can the incredibly high cost of litigation.
ask, what is a rational approach towards where I
set the alarms? For example, the heart rate Welch One more obstacle is FDA clearance for
area—is it okay to set it at 140 versus 120? any kind of rationalized improvement in
If you do that, you will get fewer alarms. The alarms. Any company that moves forward with
question is, at what point is the alarm threshold an intelligent or smart alarm solution will have
set beyond a reasonable level, such that you’re to run the FDA gauntlet of getting clearance.
causing harm? That evidence does not exist The question to the industry is, is that worth
right now. And it’s not a one-size-fits- the effort? There’s almost a resistance in the
all solution. area of alarms because it raises the specter of a
prolonged clearance cycle for any new platform
Wilcox The aircraft cockpit used to be like our that’s developed.
situation in the ICU or the operating room
(OR) today. Separate vendors made different Talley The partnership between the clinician,
devices that had alarms, and when multiple vendor, and FDA needs to be strengthened so
things started happening simultaneously, pilots we can get the national attention we need to
were overwhelmed. But they had the advantage discuss how more is not necessarily better in
of a general contractor who could create a the realm of alarms. That would lead us into a
unified, integrated system that represents an dialogue about interoperability and intelligence,
alarm philosophy that is consistent.9 and how we can use that to move ahead. There
We’re not in that is a mistaken notion that the information glut
The liability issue is a big one here. When situation because there’s puts us at an advantage; instead, we find
you talk to vendors, purchasers, trustees, no comparable general ourselves at a disadvantage.
contractor who can
and so forth, they all come back behind
rationalize everything like Hedley-Whyte Another barrier to examine is
the defense of the incredibly high cost in the aircraft industry. the Health Insurance Portability and
of litigation. And there’s no incentive Accountability Act (HIPAA) regulations about
for individual manufac- confidentiality of medical records. People are
turers. So the only way that could happen is for finally realizing that, if your survival depends
hospitals and medical facilities to demand it. on transmission of what your critical care state
There have been third-party attempts to come is, then HIPAA is a bad thing. David Pogue
up with a unified system that you plug all your wrote in Scientific American recently11 that the
devices into. To my knowledge, none of them general population of the United States needs
have gotten off the ground yet, but ultimately to be told that some regulations about confiden-
that’s going to be one of the solutions. tiality or medical care need revising, and
Congress needs to revisit the whole issue.
12 Horizons Spring
Single user license only. Copying, networking, and distribution prohibited. 2011
The Big Picture
vendor to another in a lot of parameters, it is
Logan What steps can and should healthcare very difficult for healthcare organizations to
facilities be taking to address the problems? make those decisions. Some organizations like
ECRI are beginning to test various technologies
Hedley-Whyte Solving this issue will require relative to false or nuisance alarms.
involvement from top hospital executives.
These hospital executives are generally not Hedley-Whyte When I am asked to visit other
aware of the problems we’re discussing today. medical institutions, I have found it rare that
The need for intelligent, integrated alarm hospitals have appropriate alarm policies. To
systems generally comes as a surprise to them. initiate rational alarm policy often takes a
committee or a very powerful central adminis-
Clark Individual hospital units typically follow tration. The development of rationalized
specific policies for alarm parameters, which policies and procedures should be based on
is appropriate. But at the institutional level, local evidence and published literature.
there isn’t a clear awareness of the problem. Also, the move of so much medicine and
Unfortunately, the reporting of alarm events in surgery toward home care does require urgent
The New York Times or Boston Globe that are rethinking of the alarm systems that should be
read by a trustee or hospital executive is a deployed. Nobody on a ventilator, hemodialysis
common impetus to develop an alarms machine, or infusion pump at home should be
improvement program. left without a distributed alarm system. It is
nonsense to think that a single alarm is
Wilcox Facilities can use the likelihood of false sufficient. You need a system distributed to
alarms as one of the top criteria for acquiring other rooms where other people are.
devices. Right now, that criterion is pretty low
on the list, if it’s even considered at all when Cartwright From a home healthcare
comparing devices. However, with the lack of perspective, it’s part of our assessment to verify
good metrics or research that compares one
The New Gold Standard
in Infusion Nursing
s Now available!
ursing visit www.ins1.org
Single user license only. Copying, networking, and distribution prohibited. Horizons Spring 2011 13
The Big Picture
that an alarm can be heard in every area of a because there’s no profit in it. On the acute-care
home. Desensitization to alarm sounds is also a and the long-term care side, however, industry
problem in the home environment. Because definitely can do more because that’s where
there is not a clinical person there with the monitors are today. How we solve these
patient all the time, family members come up problems is really the call to arms.
with their own interpretation as to what alarms At a children’s hospital, we analyzed what
mean because of what has happened in the past. measurements are most often alarming. We
Often, family members will actually turn off found that respiration rate was the root cause of
alarms because they’ve heard so many most of the alarms that were driving nurses,
false ones. patients, and families crazy. So that gives us an
attack point from a technology standpoint. What
Welch In fact, if you go through device can we do about improving that particular
incidents reports, you’ll find that home apnea parameter? How do we integrate all of the
monitors are one of the most egregious exam- parameters? For instance, why not couple
ples of too many false or nuisance alarms, and respiration rate with oxygenation in patients?
an inability to There are some rational ways of approaching
In 2002, the Joint Commission issued an alert on relay the true this. The question for industry is, will that
alarm safety. . . We’re almost 10 years past their alarms that provide a competitive advantage of one company
somebody versus the other?
recommendation and we’ve made little to no real
advance in solving this alarm safety problem. something Logan AAMI has just established a new standards
about. Remote annunciation of alarms is committee on alarms, because of the importance
essential because caregivers, especially ones of the issues that need to be addressed from a
outside ICU settings, are not at the bedside standards perspective. What do you think is need-
when these events occur. ed on the regulatory, accreditation, or standards
However, if you don’t solve the problems of fronts to deal with this issue?
nuisance alarms or alarm fatigue, moving those
alarms to an already-busy clinician, home Hedley-Whyte The international standards-
healthcare provider, or family member is just setting process in this area is not smooth at all.
going to annoy them, and they’re going to turn We’ve had international standards for about 40
off alarms. So first solve the alarm fatigue years on alarm systems. There’s currently a
problem, and then start thinking about how to disagreement between two standards bodies on
get alarms to the right person to rescue how to approach these systems, which needs to
that patient. be resolved.
Logan What should industry be doing to address Logan The new AAMI standards committee is
these problems? waiting to get started until they hear the needs
and priorities from all perspectives at the alarms
Hedley-Whyte There are many barriers to what summit that AAMI is co-hosting this October
industry can do to solve these problems in the 4-5 with ECRI Institute and ACCE. The commit-
home environment. In Massachusetts, we’ve tee is also waiting on a revision of an updated
had political infighting to get sophisticated, IEC foundational standard (60601-1-8), upon
distributed alarm systems deployed in the which our new work will build. If we can solve
home. You have to battle insurance companies many of the challenges with alarms from what
for reimbursement to wire another room, set up we learn at this year’s summit, then hopefully
a local area network, and so forth. this preferable non-regulatory route for
improvements throughout the system will
Welch Until there’s reimbursement for compa- eliminate the need for any additional
nies developing these body-worn sensors for regulatory action.
home healthcare, you are not going to get
venture capitalists or their companies to invest
14 Horizons Spring
Single user license only. Copying, networking, and distribution prohibited. 2011
The Big Picture
Talley In 2002, the Joint Commission issued more we can look at it from an evidence-based
an alert on alarm safety. That was helpful; it perspective, the more we can maintain the
established a call to action and gave us lever- momentum we’ve begun here.
age to say we are all accountable. We’re almost
10 years past their recommendation and we’ve Welch This is an enormously important issue
made little to no real advance in solving this in healthcare because we’re already seeing that,
alarm safety problem. The Joint Commission due to so many nuisance alarms occurring,
has since established a standard aimed at ap- hospitals are looking at not monitoring
propriate alarm settings and audible alarms. patients. Removing the technology because
It’s a start; however, it does not effectively get us there are problems with it is a step backward.
to where we need to be. Clinicians are still en- Therefore, at a stakeholder meeting, it’s going
countering alarm fatigue in the hospital setting. to be extremely important to have the right
participants representing the political, financial,
Logan What next steps are needed to solve insurance, and vendor spectrums. Everybody
these problems? recognizes that this is a worthy problem to
solve. Getting agreement and collaboration on
Talley We must continue to seek research a way forward is the only way this is going
opportunities. Healthcare providers must to be solved.
represent the patient and the family interests.
As we said earlier, vendors are looking at the Andrew-Webb With home healthcare, we need
bottom line. The providers are the ones who are to work more with vendors to make sure they’re
most closely aligned with patient/family aware of issues we are having with alarms so
interests. And it is incumbent on us to keep the we can present that to insurance companies
topic alive through research endeavors. The and try to resolve these issues together. n
1. AAMI. “Alarm Fatigue: An Issue Across the 5. American College of Clinical Engineering 9. Jacobsen A et al. Crew Station Design and
Country,” in Infusing Patients Safely: Prior- Healthcare Technology Foundation. Impact Integration, in Human Factors in Aviation,
ity Issues from the AAMI/FDA Infusion of Clinical Alarms on Patient Safety. 2006. Edition 2, edited by Eduardo Salas and Dan
Device Summit. Arlington, VA; 2010:35. Available at http://thehtf.org/White%Paper. Maurino, 2010.
2. Talley LB, principal investigator. 10. Hedley-Whyte J, ed. Operating Room and
6. Taenzer AH et al. Impact of Pulse Oximetry
“Understanding Factors Associated with Intensive Care Alarms and Information
Surveillance on Rescue Events and Intensive
Monitor Alarm Generation in Critically Ill Transfer. ASTM Special Technical Publica-
Care Unit Transfers. Anesthesiology 2010; tion (STP) 1152. American Society of Testing
Children.” IRB-approved study at Children’s
National Medical Center, externally funded and Materials, 1992.
by Philips Healthcare. January 2008.
7. Graham KC et al. Monitor Alarm Fatigue:
11. Pogue D. Don’t worry about who’s
Standardizing Use of Physiological Monitors
3. IEC 60601-1-8:2006. Medical electrical watching. Scientific American
and Decreasing Nuisance Alarms. American
equipment - Part 1-8: General requirements for 2011;304(1):32.
Journal of Critical Care 2010; 19(1):28-38.
basic safety and essential performance - Collat-
eral standard: General requirements, tests and
guidance for alarm systems in medical electrical 8. Atzema C et al. ALARMED: Adverse events
equipment and medical electrical systems. in low-risk patients with chest pain receiving
continuous electrocardiographic monitoring
4. IEC 60601-2-49:2011. Medical electrical equip- in the emergency department: A pilot study.
ment - Part 2-49: Particular requirements for American Journal of Emergency Medicine
the basic safety and essential performance of (2006); 24:62-67.
multifunction patient monitoring equipment.
Single user license only. Copying, networking, and distribution prohibited. Horizons Spring 2011 15