Alarm fatigue is a major patient safety issue in hospital ICUs, but how does the problem differ in neonatal and pediatric intensive care units?

By Bill Pruitt, MBA, RRT, CPFT, AE-C, FAARC


Alarm fatigue is a major patient safety issue in hospital ICUs, but how does the problem differ in neonatal and pediatric intensive care units? This article will explore some of the unique challenges that respiratory therapists and critical care specialists face when monitoring infants and pediatric patients, as well as some of newest technology available to assist patient safety and avoid false alarms. 

Alarms in the Spotlight

The 2020 release of the National Patient Safety Goals for hospitals from the Joint Commission includes clinical alarm systems as one of its safety concerns. According to this release, alarms are established to “alert caregivers of potential patient problems, but if they are not properly managed, they can compromise patient safety.”1 The focus on alarms in light of patient safety has a long history and has generated numerous research projects to address issues with problems like alarm fatigue where frequent alarms have overwhelmed the staff and possibly affected the response to an alarm. Almost thirty years ago a textbook from 1993 on hypoglycemia mentions alarm fatigue related to real-time blood glucose monitoring2 and a recent 2021 Google Scholar search of the term “alarm fatigue” found over 150,000 articles. 

Neonatal Intensive Care Unit (NICU) and the Pediatric Intensive Care Unit (PICU) patients have alarms established to detect a number of possible issues with their condition or possible issues with equipment being used to support or treat. Alarms are set for heart rate, heart rhythms, respiratory rate, arterial blood pressure, body temperature, and oxygen saturation levels. Equipment such as infusion pumps and mechanical ventilators also have alarms to notify issues with the patient or with the device.3 A review article on alarm fatigue from 2012 mentioned that there are about 700 physiologic monitor alarms per patient each day.4 A study from Johns Hopkins found that over a 12-day period, one ICU had an average of 771 alarms per bed per day.5 These figures explain why there is a serious concern for issues like alarm fatigue and desensitization. As the intensity of care and patient acuity increases, more equipment is found at the bedside and most of these devices have alarm packages—so the problem expands.

NICU/PICU Alarms

Alarms in the NICU and PICU cause stress to the patients, to their family members, and to the staff. Alarms may be difficult to detect due to the geographical layout of the unit(s) or in situations involving many alarms going off at the same time. Noticing a particular alarm in the midst of several alarms sounding off may be difficult, or an alarm may not be heard if there is too great a distance from the alarm and the staff. The staff may be desensitized to reacting to an alarm due to being overwhelmed by frequent alarms leading to delays in responding, ignoring, or even disabling them. These issues could lead to serious patient harm. 

A research article published in 2020 concerning strategies to reduce safely alarms mentions that pulse oximeters are one of the most common sources for alarms.3 A study reviewing alarms over a five year period (2008-2013) in a Level IIIB NICU in Boston found there were over 2.2 million alarm events from a convenience sample of over 900 patients:

  • Only 3.6% of alarms were deemed “critical,” linking to issues with serious or sudden desaturations, arrhythmias (asystole or ventricular tachycardia/ventricular fibrillation), bradycardia, or tachycardia.
  • 55.0% of the alarms were “advisory” alarms dealing with low/high SpO2, low/high heart rates, or CPAP alarms.
  • 41.4% of the alarms were “device alerts” dealing pulse plethysmograph, electrocardiogram, respiratory rate, or arterial blood pressure.
  • Pulse oximetry monitoring generated over 60% of all alarms in this study.
  • In the “critical” alarms subset, false alarms were found in 99.9% of the asystole alarms, 100% of the ventricular tachycardia/ventricular fibrillation alarms, and 35.9% of the critical desaturation alarms.6
  • There are many studies in published research that examine false alarms and a conservative estimate of data analysis has up to 90% of all alarms in the ICU environment as false positive.5 

Ventilator Alarms

Mechanical ventilators provide one of the highest levels of critical patient support and have a number of monitored settings linked to alarms to check on pressures, volumes, respiratory rate, etc. There are several problematic issues with ventilator alarms as pointed out in a 2021 editorial in Respiratory Care (RC) journal. First, machines from different manufacturers offer different alarm packages or may use different terms to define the same type of alarm. 

Second, evidence-based guidance on how to set or manage alarms does not exist so hospitals have internal policies and procedures that vary from one facility to another. Lastly, even though the Joint Commission, the Emergency Care Research Institute (ECRI), and other organizations have targeted ventilator alarms to encourage improvement in safety and effectiveness, a lack of research has hampered efforts to bring about significant change. 

This RC journal editorial also notes that in one NICU study, ventilator alarms made up 11.7% of all alarms and that another study in the PICU, some 31% of false alarms were generated by ventilator alarms.7 A study that evaluated research done on mechanical ventilation alarms found that only about 5% to 13% of the alarms were clinically relevant or described as alarms that needed action taken to resolve.8 

Efforts to Improve Alarm Safety and Reduce the Number of Alarms

Research is ongoing to use artificial intelligence (AI) in helping deal with alarms. One such project was published in 2020 where software was developed and utilized to screen for false alarms. The probability of a false alarm was established by integrating several factors including the patient movement or repositioning, battery life charge on the device generating the alarm, the last time the patient’s skin was prepped to have an electrode placed, and the last time electrodes were changed-out.5 

Another project using AI was designed to reduce the number of alarms. Although this project was experimental and worked with surgical patient cases, the approaches used to reducing alarms could be translated into the ICU environment. Software analyzed data concerning several variables including electrocardiograph, pulse oximeter, capnograph, noninvasive arterial blood pressure monitor, airway flow, and pressure monitor. In a few cases, data was included from a Y-piece spirometer, an electroencephalogram monitor, and an arterial blood pressure monitor. As alarms were triggered, the system decided whether to aggregate alarms, whether to add a label announcing the probability of it being a false alarm, and who in the team should receive notification (based on their specialization level, degree of experience, availability, geolocation, and current workload conditions). The system was designed to notify the caregiver of the initial alarm but hold subsequent alarms concerning the same issue if the subsequent alarms occurred within a 5-minute interval. The outcome of this project was a reduction in notifications for up to 99.3% of the total of all alarms for a given staff member and a 99.17% reduction in the number of total alarms in the unit.9 

Other projects have looked at how alarms are set in terms of alarm limits and alarm delay. Mayo Clinic in Rochester, MN worked on using consecutive cycles of Plan-Do-Study-Act (PDSA) to reduce the alarm burden in their Level IV NICU by addressing alarm limits, alarm delay, and utilizing a new pulse oximeter sensor. This project gathered and analyzed initial data from their vital signs monitors. After reviewing the data and brainstorming within a Quality Improvement team, they found that between the heart rate, respiratory rate and SpO2 monitoring, the SpO2 alarms for saturation were most frequent. Within the SpO2 alarms, 6% were categorized as critical and 94% were categorized as advisory. Within the advisory SpO2 alarms, the staff silenced 33% while 67% were self-resolving alarms. Further investigation found that 98% of the self-resolving alarms required no intervention (2% resolved after an action was taken, ie increasing FiO2 or repositioning the patient). 

Through a series of PDSA cycles changes were made in the pulse oximeter alarm limits from a low-high of 90%-95% to 86%-97%, a new pulse oximeter sensor was evaluated, and the time delay for low and high saturation alarms were increased from 10 to 20 seconds. After each PDSA cycle, 24 hours of data was gathered to check on the outcomes of the changes. The final results of these actions saw a decrease in the mean number of self-resolving SpO2 alarms per hour from 14 to 6, a 64% reduction. Throughout the PDSA cycles, percent time spent in desired oximetry target range did not change and the researchers found that the new oximeter probe resulted in more consistent and more accurate SpO2 readings, thus also contributing to the reduction in false alarms. In a pre-post project survey of the staff, the results showed that parents’ concerns about the promptness of staff response showed significant improvement.10

Another study that adjusted the lower alarm limits and increased the time delays on their pulse oximeters resulted in about a 78% reduction in the mean number of self-resolving low alarms.11

Improvements have been made in mechanical ventilation alarms as manufacturers have included an alarm autoset feature that allows thresholds to be established automatically by pushing a button. For example, once the patient-ventilator system has stabilized after initiation of mechanical ventilation, the autoset feature will set high/low alarms for minute volume at +30% of the current system status. The concept of “smart alarms” in ventilators is evolving. One example are alarms that will set to certain defaults levels based on clinician inputs of patient category (neonate, pediatric, adult), patient weight, and mode of ventilation. Smart alarms incorporate escalating priority by changing the visual alert (ie, yellow changing to red), the type of tone generated (ie, two beeps that drop the notes changing to four beeps that move up the tonal scale), and/or increasing the volume of the alarm if it is not resolved or silenced within a certain time. Coupling alarms (ie, low tidal volume and low pressure alarms) can help identify a high priority “patient disconnect alarm” versus more conventional stand-alone alarms for low volume and low pressure. Another smart alarm is found in the alarm silence button for suctioning, which pauses the alarm for a short time while also pre-oxygenating the patient for this procedure then resets to the previous point at the end of the time period.12

Making a Difference

Efforts to make changes in the number of alarms and improve the trust in the alarm monitoring systems need to have good data to establish a baseline and objectively measure outcomes. There needs to be a team approach to the problems of alarm fatigue and desensitization that involves nurses, respiratory therapists, physicians, biomedical engineers, and administration at the minimum. Thorough education of the bedside team and unit support staff is crucial to ensure that all are aware of the problems, the changes that are being made to resolve issues, and have a good understanding of the alarms themselves. The team needs to know the significance of alerts, how the alarms function, and the policies/procedures that come out of a project of this magnitude. One excellent resource for starting an alarm management project comes from the Advancement of Medical Instrumentation (AAMI) Foundation’s National Coalition for Alarm Management Safety and their document “SpO2 Alarm Management Toolkit” which can walk a team through the process step-by-step.13 

Conclusion

Alarms are a valuable tool in providing safe, effective support and therapy to critically ill patients. However, alarm fatigue and alarm desensitization is a very real problem with serious consequences. Efforts to improve the NICU and PICU environment and reduce/refine alarms are making progress but more needs to be done to improve and manage the technology and the human behaviors to bring significant change. Healthcare workers need to know what is being done and how they can contribute to these efforts.


RT

Bill Pruitt, MBA, RRT, CPFT, AE-C, FAARC, is a writer, lecturer, and consultant. He has over 40 years of experience in respiratory care, including more than 20 years teaching at the University of South Alabama. Now retired from teaching, he continues to write and provide guest lectures. For more information, contact [email protected]



References

  1. Joint Commission National Patient Safety Goals: https://www.jointcommission.org/-/media/tjc/documents/standards/national-patient-safety-goals/2020/npsg_chapter_hap_jul2020.pdf
  2. Thow JC, Home PD. Nocturnal hypoglycaemia. Hypoglycaemia and diabetes. Edward Arnold, Sevenoaks, UK. 1993:284-301.
  3. Varisco, G., van de Mortel, H., Cabrera‐Quiros, L., Atallah, L., Hueske‐Kraus, D., et al. 2021. Optimisation of clinical workflow and monitor settings safely reduces alarms in the NICU. Acta Paediatrica, 110(4), pp.1141-1150.
  4. Cvach M. Monitor alarm fatigue: an integrative review. Biomedical instrumentation & technology. 2012;46(4):268-77.
  5. Fernandes C, Miles S, Lucena CJ. Detecting false alarms by analyzing alarm-context information: algorithm development and validation. JMIR Medical Informatics. 2020 May 20;8(5):e15407.
  6. Li T, Matsushima M, Timpson W, Young S, Miedema D, Gupta M, Heldt T. Epidemiology of patient monitoring alarms in the neonatal intensive care unit. Journal of Perinatology. 2018 Aug;38(8):1030-8.
  7. Ruppel H, Funk M, Whittemore R. Measurement of physiological monitor alarm accuracy and clinical relevance in intensive care units. Am J Crit Care 2018;27(1):11-21.
  8. Scott JB. The Conundrum of Mechanical Ventilation Alarms. Respiratory Care 2021 April; 66 (4) 699-700.
  9. Fernandes CO, Miles S, De Lucena CJ, Cowan D. Artificial intelligence technologies for coping with alarm fatigue in hospital environments because of sensory overload: algorithm development and validation. Journal of medical Internet research. 2019;21(11):e15406.
  10. McCauley KE, Schroeder AA, DeBoth TK, Wiebe AM, Bosley CL, Ballweg DD, Fang JL. Reducing Alarm Burden in a Level IV Neonatal Intensive Care Unit. Pediatric Quality & Safety. 2021 Mar 1;6(2):e386
  11. Johnson KR, Hagadorn JI, Sink DW. Reducing alarm fatigue in two neonatal intensive care units through a Quality Improvement Collaboration. Am J Perinatol. 2018;35:1311–1318.
  12. Walsh BK, Waugh JB. Alarm strategies and surveillance for mechanical ventilation. Respiratory Care. 2020 Jun 1;65(6):820-31.
  13. AAMI SpO2 Alarm Management Toolkit: https://www.aami.org/docs/default-source/foundation/alarms/aamifdn_2017_spo2_toolkit.pdf?sfvrsn=ca90c27d_2. Accessed 9/21/21.