Three Pillars of Clinical Alarm Safety
A simple framework addresses the problem of chronic alarm fatigue
in three areas, in a logical sequence.
By Tim Gee
Management of medical device alarms has been a persistent challenge for decades (ECRI Institute, 1974). Histories of surveys, papers, and other initiatives to improve alarm safety have been compiled (Clark, 2005; ACCE Healthcare Technology Foundation, 2006; ECRI Institute, 2008), and yet the problem persists. Clinical alarm safety remains a problem, not because clinicians and caregivers don’t care, but because the best practice for clinical alarm management is wide-ranging and difficult to achieve. And once optimal clinical alarm safety is achieved, constant vigilance and effort is required to sustain and improve best practice over time.
Clinical alerts and alarms indicate an immediate safety risk to the patient due to a variety of hazardous conditions or the patient’s deteriorating clinical condition. These clinical alarms have traditionally been generated by medical devices but increasingly can be generated by clinical decision support systems and other information systems. These new sources of clinical alarms provide monitoring and analysis of near real-time physiological and therapy data from patient-attached devices along with patient data from a variety of information systems. Clinical alarms differ from more generalized alarm safety in health care by their ability to indicate immediate patient risk. More generalized alerts and alarms, such as warnings generated by EMRs for potential negative drug interactions or contraindicated therapies represent a different, less immediate level of patient safety risk.
The most common symptom of clinical alarm safety problems is chronic alarm fatigue, a pervasive problem in many hospitals. Alarm fatigue occurs when users become desensitized to alarm signals due to over stimulation from alarms. Alarm fatigue results from exposure to too many patient alarms and the generation of false positive or otherwise non-actionable alarms that overwhelm caregivers’ ability to perceive and respond to true positive alarms. For a discussion of alarm classifications, see “Alarm Fatigue Hazards: The Sirens Are Calling” (Welch, 2012).
When reviewing the literature on clinical alarm safety, many improvement efforts offer scatter shots of recommendations or best practices. The subject is so broad that it is very difficult to present a comprehensive overview. Addressing alarm safety issues often seems disjointed because of the apparent absence of a simple framework to categorize and structure a comprehensive approach to framing the issues impacting clinical alarm safety. This article is an attempt to put forward just such a framework: three pillars of clinical alarm safety:
- the clinical practice of alarm safety,
- alarm notification, and
- reporting and analytics.
The key components of each pillar are described, noting recommendations and best practice when applicable.
Clinical Practice
The clinical practice of alarms embodies most of the immediate generation of alarms. The clinical practice of alarms is where most false positive and non-actionable alarms are reduced or eliminated. This clinical practice is critical regardless of any technology used for alarm notification. The goals of this clinical practice are to minimize the volume of alarms generated by reducing false positive and non-actionable alarms.
This pillar starts with the proper selection of monitoring device and physiological parameters for the patient. While this seems self evident, this can be a challenge in certain cases. For example, a substantial percentage of telemetry packs are purchased for use on patients at risk of a deteriorating clinical condition rather than arrhythmias. These at-risk patients are a challenge because their on-service units often lack patient monitors, and traditional high acuity units with multi-parameter patient monitors provide a level of care and cost that exceeds their clinical need. For many years, the telemetry pack has been the small, portable, and easily tolerated patient monitor applied to low acuity patients. Telemetry packs tend to do a poor job of early detection of a deteriorating clinical condition and can also generate frequent false-positive and non-actionable alarms, contributing to alarm fatigue. This example highlights the need to identify and assess all patient populations that generate clinical alarms, ensuring that optimal equipment, parameter, and alarm limit choices are made to ensure safety while minimizing the false positive and non-actionable alarms that contribute to alarm fatigue.
Hospitals are under tremendous financial pressures, resulting in a constant search for ways to minimize the costs of equipment, consumables, and supplies. The choice of sensors for monitoring should balance cost against factors impacting alarm fatigue. Poor quality sensors can be hard to place on the patient and easy to dislodge. Insufficient quality can mean that sensors are unreliable sources for physiological data, which results in false positive alarms. Sensors that generate higher than normal levels of leads-off or false-positive alarms should be replaced.
Even the logistics of receiving sensors from distributors, moving them to nursing units, and making them available for use can impact alarm fatigue. Some sensors require certain environmental conditions for proper storage and can become degraded over time. Inventory management procedures and storage locations should be reviewed to ensure sensors do not exceed their expiration dates and are used when fresh and at peak quality. Likewise, the logistics and storage of sensors must be sufficiently convenient to remove the need for possibly counterproductive workarounds by members of the staff.
To acquire signals with maximum fidelity and minimum artifact, sensors must be properly placed on the patient. This can entail skin prep, optimal sensor selection for the best sensor site on the body (e.g., ear, forehead, finger tip, etc.), and thoughtful adjustment and placement of cables or tubes connecting the patient to the medical device. Once sensors are optimally placed, nurses must be vigilant to ensure that sensors are properly maintained and replaced when needed, in order to maintain signal fidelity and minimize false positive alarms.
A more complex element of alarm clinical practice is the optimization of default alarm parameters and limits. Determining the optimal physiological parameters and alarm limits for specific patient profiles is a time-consuming process. Often a committee or task force made up of clinicians and caregivers must reach consensus regarding default parameters and alarm limits. These are issues on which reasonable people can differ, in spite of looking to the literature for guidance. Best practice entails reviewing the data on alarms generated within the relevant unit. This alarm data provides quantitative data that can indicate alarm frequency by parameter and show the associated physiological values to help optimize alarm limit defaults. Patient monitoring systems that include networked medical devices and a central station have the ability to report this kind of alarm data. The ease or difficulty of access to historical alarm data varies by manufacturer and the age and type of medical device system. Some effort may be required to format the data into something the task force can use for analysis.
Default alarm parameters and limits are the starting points for alarm generation and work well for most patients. The occasional patient, due to unique physiology or condition, will produce an exceptional number of false-positive or non-actionable alarms. Consequently, it is important that procedures exist to facilitate the adjustment of default alarm parameters and limits for these patients. These procedures must be sufficiently flexible and efficient so that revising the defaults for a patient can be one before the effects of alarm fatigue are manifested. Any procedure that precludes revising default values for a period of hours is not practical and could put all patients on the impacted unit at risk from alarm fatigue.
The foregoing activities represent the clinical practice of alarms. Taken together these tasks and considerations represent essential activities necessary to minimize false-positive and non-actionable alarms that most contribute to alarm fatigue. Once an effective clinical practice is put into place, the next pillar, alarm notification, must be considered.
Alarm Notification
The alarm notification pillar encompasses the means by which alarms are annunciated and reliably conveyed to the appropriate caregiver. Medical devices are designed to provide alarm notification within the close proximity of the medical device as part of their basic function. Information system technologies are also available to acquire alarm signals, route them to a designated caregiver regardless of location, and ensure an alarm receives an appropriate response. Whether the audible and visual alarm notification generated by the medical device or a more sophisticated alarm notification solution is required depends on the physical layout of the nursing unit and the ratio of patients to caregivers.
Conventional, standalone medical devices are intended to provide alarm notification in situations where caregivers can readily hear and see alarm indicators. Some care delivery areas are ideal for conventional medical device alarm notification capabilities. An example is a traditional post-anesthesia care unit (PACU) where patients are arranged around the perimeter of a mid- to large-sized room. Often in these units there is a desk or nursing station in the center of the room where staff members work when not at the bedside. In this scenario, caregivers can easily hear alarm annunciations and see visual indicators such as blinking displays or flashing colored lights on the medical device from anywhere in the unit. In this environment, alarm volumes can be kept at a moderate level, and alarm conditions can be easily discerned and responded to in a timely fashion. Acoustic treatments of the room’s interior can also enhance conventional medical device alarm annunciation.
Likewise, the ratio of patients to caregivers is a factor in determining the suitability of alarm annunciations that come directly from the medical device. Patient care areas where there is a one-to-one or two-to-one ratio of patients to caregivers can be well served by conventional alarm annunciation. For example, an intensive care unit (ICU) organized into pods that contain two patients in each pod are often well served by alarm annunciations produced by the medical device. Again, proper acoustic treatment of the ICU’s interior space is recommended.
Conventional alarm notification starts to break down when patients are widely dispersed and out of view of caregivers. This happens when patients are placed in private rooms. Add to this a higher patient to caregiver ratio, often around eight patients to one caregiver, and the alarm annunciations generated by medical devices is either insufficient to be heard in all areas of the unit or so loud that everyone on the unit—all staff members, patients, and their families—are inundated by a cacophony of alarm signals. The former condition results in missed alarms and the latter results in alarm fatigue; patient safety is compromised by either situation.
One of the root problems of alarm fatigue or missed alarms is that medical devices that were intended to be used within easy audible and visual range of the caregiver are now used in nursing units where patients are isolated from one another and only occasionally seen by caregivers. The conventional solution to this problem is to extend the alarm capabilities of the medical devices by routing audible alarm signals through public address (PA) systems and displaying lights or visual alarm messages using hallway ceiling lights, remote message panels, or remote surveillance displays. This approach certainly preempts missed alarms, as every alarm on the nursing unit is audible throughout the nursing unit, often with accompanying visual information via remote message panels, etc. Unfortunately, a concomitant effect of this “broadcasting” of all alarms throughout a unit is alarm fatigue.
Alarm notification systems, or messaging middleware, are information technology systems that acquire messages (e.g., clinical alarms), determine the appropriate recipient of the message, and then ensure a response. A response is ensured through closed-loop communications, where the system tracks whether or when a message was received, read and, at times, accepted as a task by the recipient. Thus clinical alarms are initially routed to the caregiver assigned to the patient generating the alarm. Should that caregiver be busy or unavailable, the system automatically reroutes the clinical alarm to a backup caregiver. Clinical alarms can be rerouted or escalated a number of times.
Advanced alarm notification systems can also improve alarm response. Certain clinical alarms are susceptible to generating false positive alarms, often a consequence of motion artifact. For example, displaying a snippet of waveform associated with the clinical alarm allows caregivers to identify false positive alarms at a glance, rather than running to the patient room to look at the medical device display. More generalized alarm context, such as certain lab values or other diagnostic or therapeutic data, can also inform caregivers when a clinical alarm is delivered, allowing the caregiver to rapidly formulate an optimal response to the clinical alarm.
A common type of non-actionable clinical alarm is the transient alarm, where alarm limits are exceeded for a few seconds at a time and then return to normal. Medical devices and early alarm notification systems annunciate or deliver those alarms to caregivers, regardless of duration. Current alarm notification systems should remove an alarm message from the user’s message queue when the alarm condition ceases to exist. An advanced alarm notification feature goes one step beyond this to provide a configurable time delay that only sends the clinical alarm message to the caregiver when the configurable delay is exceeded. This delay feature allows alarm notification systems to filter out transient alarms that have been determined by the hospital to be non-actionable. As advanced alarm notification systems adopt clinical decision support system features, it becomes possible to identify and filter duplicate alarms—different parameters from the same medical device, or different medical devices attached to the same patient—so that caregivers are not barraged by duplicate clinical alarms.
The alarm notification system also logs all clinical alarms and other messages generated, the routing for each alarm, timing of responses, and the final disposition of the clinical alarm, providing excellent operational information.
Reporting and Analytics
Minimal reporting and analytics are provided by log files from patient monitoring and other medical device system servers. As noted above, this data is analyzed to facilitate the optimization of default alarm parameters and alarm limits. For best results, this log data should be transferred to an analytics tool with capabilities like sorting, trending, and the display distribution curves and other basic analytic tools. Basic data indicates the parameter alarming; the start, stop, and length of the alarm condition based on the alarm limits; and the actual physiological data that was produced by each individual parameter during the alarm condition. This basic data may also include the patient or the room number that generated the alarm.
An alarm notification system takes this basic medical device alarm data and adds additional context. This additional information includes patient-to-staff assignments, routing information of alarms, acceptance or declining of the alarm message, escalation of the alarm to another user, and final disposition. Taken together, this alarm information can be very valuable in managing patient care units.
The initial reporting and analytics application for alarm notification systems is a tool for improving the clinical practice of alarms, and especially in optimizing default clinical alarm parameters and limits. In cases where the design of the unit does not necessitate use of an alarm notification system, the system should be used to capture alarm data in that unit for the reporting and analytics features alone.
Alarm notification systems can also serve as forensic databases when investigating sentinel events. The value of this capability increases as additional systems, such as nurse call, patient beds, and other medical devices are integrated with the alarm notification system. This database provides a detailed sequence of events that could start with a bed exit alarm and all subsequent responses in the case of a patient fall. Like any other patient safety investigation, the intent is to gather a detailed picture of the sequence of events to identify systemic problems for remediation.
The operational and clinical data collected by alarm notification systems can also be put to use improving operations. The data collected can serve as indicators of staff training requirements, caregiver workload, and even provide insight into patient acuity. The adage that something must be measured before it can be managed applies with alarm notification systems. There is a whole new realm of operational and clinical data waiting to be mined to improve HCAHP scores, patient outcomes, and staff productivity.
These three pillars of alarm notification provide a simple framework for tackling the problem of chronic alarm fatigue. And while it is not a detailed roadmap or project plan, the pillars divide the scope and areas of focus for alarm notification into a logical sequence. This framework should also be of some value for addressing the Joint Commission’s new National Patient Safety Goal on alarm management (2013).
Tim Gee is principal and founder of Medical Connectivity Consulting (www.medicalconnectivity.com) where his practice revolves around workflow automation through the integration of medical devices with information systems and enabling technologies. Gee is the 2012 recipient of the American College of Clinical Engineering’s Challenge Award and is program chair and principal organizer for the Medical Device Connectivity conference. Gee writes at the Medical Connectivity blog, serves on the editorial advisory boards of a number of publications including Patient Safety & Quality Healthcare, and participates in key industry interoperability development efforts. He may be contacted at tim@medicalconnectivity.com.
Gee, T. (2014). Three pillars of clinical alarm safety. Patient Safety & Quality Healthcare, 11(2), 46–51.
References
ACCE Healthcare Technology Foundation. (2006). Impact of clinical alarms on patient safety. Plymouth Meeting, PA: author. Retrieved from http://thehtf.org/white%20paper.pdf
Clark, T. (2005, September/October). Clinical alarms remain an area for patient safety improvement. Biomedical Instrumentation & Technology, 357–358.
ECRI Institute. (1974, July). Hazard: Gaymar hypothermia machine. Health Devices, 229-230.
ECRI Institute. (2008, May). Clinical alarms. Healthcare Risk Control, 4. Retrieved from https://www.ecri.org/Documents/RM/HRC_TOC/HRC_CriticalCare_ClinicalAlarms.pdf
The Joint Commission. (2013, December 11). Alarm system safety. R3 Report: Requirement, Rationale, Reference, 5. Retrieved from http://www.jointcommission.org/assets/1/18/R3_Report_Issue_5_12_2_13_Final.pdf
Welch, J. (2012). Alarm fatigue hazards: The sirens are calling. Patient Safety & Quality Healthcare, 9(3), 26–33. Retrieved from http://www.psqh.com/mayjune-2012/1291-alarm-fatigue-hazards-the-sirens-are-calling.html