Improving Alarm Management Practices

Abstract

This study reviewed the application of sociotechnical models and frameworks to reduce wireless calls without introducing risk and impacting patient care, supplementing the findings of the study conducted by Clodfelter (Studies in Health Technology and Informatics 2024;315:463–467). This study was conducted at an 815-bed Magnet-recognized facility with comprehensive trauma, children's, and stroke services. Two models, both by Sittig and Singh, were applied to evaluate interdependent and interrelated concepts of human and technical components in the sociotechnical work system of medical-surgical nursing units. Sittig and Singh's (2010) eight-dimensional sociotechnical model comprehensively represents the factors influencing the design, development, use, implementation, and evaluation of health information technology. Interventions were piloted on three medical-surgical units with like hardware and software wireless integrations. The quantitative analysis demonstrated the effectiveness of the interventions, showing significant reductions in nuisance alerts without introducing new risks. Furthermore, there were no unintended consequences identified following the implementation of the intervention. There are direct, qualitative benefits related to decreased nuisance alarms and clinician experience. Return on investment and the value proposition reflect both tangible and intangible costs. Workflow changes, policy revisions, and system data were used to demonstrate meaningful improvement without unintended consequences. Tailoring workflows by addressing sociotechnical factors can reduce nuisance alerts and improve usefulness of systems.

Nurses interact with a multitude of medical devices daily to care for patients. Clinical equipment can generate true and false alarms.¹ Nonactionable alarms are true signals but do not require action by the nurse because they are not clinically relevant. False alarms are typically a result of a process or system issue. For example, a faulty sensor or incorrect use of the equipment triggers the signal.^2,3 Avoidable false and nonactionable alarms are referred to as “nuisance alarms” and are noted as the highest contributor to alarm-related events.^1,4–6 Alarm management is the process and interventions used to reduce nuisance alarms and improve meaningful alarms to prevent alarm fatigue.^1,3,7 Alarm fatigue is a widely known phenomenon when a nurse becomes desensitized by excessive signals. This overload of alarms and alerts causes delayed response because the nurse finds it hard to prioritize action or simply stops responding because of questionable system reliability, either of which could lead to safety issues.^1,3,7–10

BACKGROUND AND INDUSTRY PRACTICES

Multiple national organizations have escalated growing issues with alarm-equipped systems. In 2013, The Joint Commission published a Sentinel Event Alert due to patient deaths and injuries related to alarm-related issues.⁸ The Joint Commission defines a sentinel event as any “patient safety event that results in death, permanent harm, or severe temporary harm” and impacts both patients and caregivers.¹¹ Some of the contributing factors were alarm fatigue, lack of customized settings, lack of staff training, inadequate staffing resources to respond, and lack of integrated devices.⁸ In 2014, The Joint Commission created the National Patient Safety Goal NPSG 06.01.01, which prioritized clinical alarm management and safety practices to reduce harm.^1,11 The Emergency Care Research Institute is a nonprofit, independent organization and designated evidence-based practice center by the US Agency for Healthcare Research and Quality. Since 2007, it has regularly reported clinical alarm systems in its list of “Top 10 Health Technology Hazards.” ^1,7,8,12,13

The integration of a call-light system to send signals directly to wireless devices, such as smartphones carried by the nursing staff, if a patient attempts to get out of bed unassisted was ascertained. Integrating smartphone technology enhanced patient safety by supporting nurses' ability to communicate and prioritize care delivery directly.^14–17 Burkoski et al¹⁵ revealed that nurses view the integration of the smartphone with the patient call-light system as integral to delivering safe care. This ability to integrate with mobile devices and appropriately escalate events is even more relevant to nurse call systems because it is the “last refuge for help, as well as the first line of defense in accident prevention.”¹⁸ There has been less focus in the literature on the nurse call-light system's impact on the environment of care and evidence-based practices for wireless integration, as compared with other devices.^15,19

MATERIALS AND METHODS

Setting and Project Design

The healthcare organization identified is devoted to utilizing innovative technologies that foster patient care. This 815-bed Magnet-recognized facility with comprehensive trauma, children's, and stroke services has implemented a HIPAA-compliant clinical communication and collaboration platform over reliable wireless networks and smartphone technology. This infrastructure integrates multiple health information technologies (HITs) such as nurse call lights, electronic health records (EHRs), and telemetry monitors to send wireless calls and alerts to care teams. During prework analysis, three medical-surgical units with similar bed counts were found to have unique monthly trends of average wireless alerts. Pilot units A, B, and C had 38, 32, and 40 beds and could be rated low, medium, and high for alert counts, respectively. All units maintain full patient capacity. Average nursing productivity was also compared to ensure no obvious staffing differences between units by organizational standards.

By being closely engaged with the implementation of the clinical communication and collaboration platform and its nurse call-light integration, the corporate nursing informatics team was able to quickly identify wireless nurse call events as a potential source of signal burden and optimization. Clinical and information technology (IT) stakeholders at the facility, division, and corporate levels participated in onsite workflow observations, interviews, policy reviews, and data analysis. Several years prior, the call-light system had limited or no wireless devices and/or smartphones, so the hospital task force saw it as an exercise to optimize the balance of having a hospital-wide communication and collaboration platform while reducing redundancy in information and processes. Establishing a reliable and ongoing alarm management process requires end-user buy-in.^1,20,21

Model and Framework

Two models, both by Sittig and Singh,²² were applied to evaluate interdependent and interrelated concepts of human and technical components in the sociotechnical work system of medical-surgical nursing units. Improving alarm safety requires addressing technical components such as changing alarm thresholds and settings, as well as human components such as addressing staff training or inconsistent staff responses.⁴ Sociotechnical models account for evaluating and optimizing HIT interventions in the fast-paced, high-pressure nursing environment.

Several components are involved in the flow of data and information between beds, nurse call-light systems, and smartphone technologies. First, the patient beds are high-tech with built-in graphical displays. The nurses must learn to navigate the touch screens to weigh the patient and arm the bed exit system when a patient is deemed at risk for falls. Next, for the alarm to reach the nurse's phone, the bed must be plugged into the wall and transmit the signals through the nurse call system. This hardwired integration allows the wireless signal to reach the correct caregiver and smartphone once the bed exit is triggered. When the alert sounds on the phone and in the surrounding corridor, the nurse can respond immediately in the room and/or act on her phone. For urgent events such as fall risk, staff must enter the room and cancel the signals to confirm a response. For a bed exit alarm, this cancelation includes the signal from the bed and the call-light hardware; otherwise, the hallway annunciations, dome lights, and phone alerts continue to repeat. Table 1 summarizes the eight-dimensional sociotechnical factors.

Table 1.

Sociotechnical Factors Applied to Wireless Alerts

Sociotechnical Factors	Description
Workflow and communication	• It represents the processes to ensure patient care is carried out effectively, efficiently, and safely. • The staff needs to understand response expectations and what the escalation process is if one is far away from the patient or busy with another patient.
Human-computer interface	• It is the aspects of the devices and software that support the interaction of the user. • This could include the color of lights from devices, the tones from the phone, or the complex buttons on the bed to enable and disable the bed exit itself when getting a patient in and out of bed.
People dimension	• It is inclusive of everyone interacting with healthcare technology to deliver patient care. • This includes not only clinical staff, but also the IT developers, informaticists, and nursing leaders who need to ensure the appropriateness of the design, policy, and implementation. • This could even include the family members getting that patient up to the bathroom and triggering a false alarm to the staff.
External rules, regulations, and pressure	• These are external forces or constraints; Food and Drug Administration regulation of alarms and devices continues to evolve, which impacts health systems and manufacturers on the design of signal pathways and decision support tools. • Other external pressures include nurse shortages that shift patient ratios and stress the work system.
Hardware and software	• This is the purely technical dimension that comprised all physical components to keep applications running, including wireless networks and bed integration capabilities.
Measurement and monitoring	• This looks at the system's availability, use by stakeholders, effectiveness, and associated consequences in terms of both intended and unintended. • It is possible to evaluate the number of bed exit alarms and alerts and staff performance, as well as to audit other reports such as IT help desk logs for device issues, including broken beds and phones, and even user access.
Internal culture, policies, and organizational structures	• These influence the other dimensions. Without leadership support, approval of policy changes or capital spending would not be possible. • Local clinical culture sustains the practice of workflow and policy.
Clinical content	• This factor centers on the data, information, and knowledge stored in a system. • The room number and event type, for example, lead to the information generated on the phone or other consoles on the unit to help caregivers make rapid decisions.

The HIT framework follows the principles of continuous quality improvement to measure, monitor, and improve events that compromise the sociotechnical work system.²² Singh and Sittig adapted this from the Office of the National Coordinator for Health Information Technology's SAFER guidelines for EHRs, which lists integrated clinical communication as one of its nine foundational groups.²³

Description of the Intervention

Studies have noted that targeting immediate care team members could improve response and that some calls may not need to be received on the wireless device.^16,24 Although most interventions were addressed through policy revision and education, rerouting wireless calls to a subset of staff required coordination and funding approval with the vendor, hospital, and corporate product engineer. The project management plan incorporated interventions, including education and escalation changes, as approved by the hospital and division clinical and IT leaders. To prepare staff for the upcoming pilot, education was assigned to nursing staff 30 days in advance of the go-live date with new workflow expectations and vendor escalation changes. Software configuration changes were tested in the corporate laboratory environment before go-live.

Multiple processes were standardized in the end-to-end bed exit workflow. For the first subprocess of staff assignment and device management, this included setting expectations that nurse call assignments and secure smartphone logins occurred within 30 minutes of shift change. This education campaign surrounded other smartphone best practices including docking devices during lunch to allow more charging time, checking device volume, and managing smartphone actions to avoid repetitious alerts that create added noise.¹⁵ Device availability impacts users and subsequent alerts. However, inventory practices were given low priority as stakeholders considered it a unit-driven problem and process.

Nurses and patient care technicians were trained on the proper procedures for responding to a triggered bed exit alarm. This comprehensive education ensured that all staff members were equipped to manage bed exit alarms and prioritize patient safety effectively. Handling bed alarms appropriately is a key factor in preventing false alarms from triggering during normal patient care activities such as bathing, turning, or dangling a patient at the bedside during physical therapy. Staff turnover and multiple bed models compromise the success and sustainment of this intervention. To reinforce unit-based responsibility in a complex environment, the education campaign included a reiteration of the previously established “no pass zone” that requires all staff, clinical and nonclinical, to address an active patient alarm or call light when passing by the door. Standardized practices were identified, using the bed alarm for high-risk patients, and the “no pass zone” policy.

RESULTS

Outcomes of the Implementation of the Intervention

System reports can be pulled from the nurse call system on bed exits triggered and response times. System data are also available on the number of wireless alerts generated. These wireless alerts are generated when the patient attempts to get out of bed and triggers the bed exit alarm. The wireless alerts can also be detailed by action taken on the phone: either click “Accept,” click “Decline,” or take no action, which is documented as “Missed.” Some factors cause variations in bed alarms and wireless alerts. The number of bed exit alarms enabled at any given time is based on the number of patients at risk for falls and the staff compliance with its use, or misuse. The number of wireless alerts is influenced by the number of bed alarms triggered on the bed, which can be a true or false event, and the number of staff programmed and on duty to receive the smartphone alert. Each bed alarm triggered can generate many wireless alerts, and both have expected variations. Only the total counts for each system are known. To summarize, the bed exit alarm has a one-to-many relationship with wireless alerts, and several factors generate variations in wireless alert counts.

During preliminary analysis of the medical-surgical units, bed exits consistently generated the highest category of smartphone alerts from the nurse call system. Table 2 shows the total counts for all wireless (smartphone) nurse call-light alert types versus only bed exit alerts. For the month of November, the pilot units had a combined average of almost 50% bed exit alerts.²⁵

Table 2.

Initial Analysis: Wireless Bed Exit Alerts Compared With All Nurse Call Alerts

Month	Unit	Total Wireless	Bed Exit	Other Events	% Bed Exit
Nov	A	24 194	16 530	7664	68
Nov	B	26 293	11 666	14 627	44
Nov	C	51 818	18 895	32 923	36

The data revealed that most alerts had no action taken on the phone, and the nursing staff determined that not every alert was actionable for every staff member. During the observations and interviews, many staff, including bedside nurses, care technicians, and unit charge nurses, noted cues such as hallway lights and tones for prioritizing responses to bed exits. Staff used the wireless notifications as an adjunct if in a closed room and unable to hear the other signals—or, less frequently, as an audit log to support escalation of patient workload and potential sitter needs. The wireless notifications were shifted only to the assigned care team of primary nurse, patient care technician, and charge nurse instead of all nursing staff on duty for that shift.

Quantitative Results Based on the Framework Outcome

Bed Exit Alarm and Wireless Alerts Findings

Figure 1 shows that under the nurse call-light system, the bed exit events are related to the number of wireless smartphone alerts received for each unit prior to and following go-live. Wireless alerts were reduced by approximately 75% in each unit.²⁵ Escalation changes stopped nonactionable alerts for staff who would not be in proximity. This was especially impactful for larger units with more beds. Nurses who float to other units and not appropriately taken off duty on another unit were also impacted here. Also seen in Figure 1 is a decrease in the number of bed alarms triggered. Unit A had the most substantial change. This could be related to the standard bed management settings introduced with education or coincidental utilization by at-risk patients. Either way, this did not add risk to the system and did reduce burden to some extent.

FIGURE 1.

Bed exit events and wireless calls, preimplementation/postimplementation.

Smartphone Actions Findings

Figure 2 shows a comparison of smartphone actions for all the bed exit alerts across each unit, prior to and following go-live. Missed calls did not fluctuate more than 10%, which mirrored prebaseline data, although this represents fewer alerts postimplementation. As a result of the analysis and interventions to decrease nuisance alerts, the summary data reflect fewer declines because fewer unnecessary alerts are being sent. Accepted calls remain the most variable and represent the smallest count of user events in the total system. As a reminder, staff must take action in the patient room to end the alert pathway so the phone actions often become an afterthought to their immediate patient needs. Therefore, addressing the multiple factors identified is crucial for improving adherence to smartphone actions and enhancing overall patient safety in the healthcare setting.

FIGURE 2.

Smartphone alert actions, preimplementation/postimplementation.

Correct Use of Bed Exit Alarms and Wireless Alerts

If cancellation within 10 seconds is used as a proxy for false events, more than 30% of the bed alarms could be indicative of a bad process that triggers a false alarm to all the other staff and increases the amount of irrelevant information coming to the phone. Therefore, for ease of comparison, there was a fluctuation in percentage of false alarms across all units before go-live. After implementation, there was not a considerable reduction in the false alarms. Ancillary staff contribute to the generation of false alarms during patient therapy and were not targeted as an entire department during this limited pilot. These house-wide services may need focused education postpilot.

Safety Improvement Processes

The third domain of the Health Information Technology Safety (HITS) framework focuses on the safe monitoring of patient events and concerns. The analysis results revealed that no unintended consequences were observed as a result of the implementation of the intervention. Average response times for the canceling of the bed exit alarm in the room averaged 19 to 42 seconds and did not increase after the pilot. By going from all-hands to user-relevant calls, there was not only a massive drop in total wireless calls but also no increase in response times, as shown in Figure 3. Fall rates on individual units were monitored through standard reporting software and showed no sudden increase or new trends over the course of 4 months before go-live and 4 months after go-live.

FIGURE 3.

Response time for bed exit alarms, preimplementation/postimplementation.

Qualitative Results

Qualitative data were collected from nursing roles to provide further information about stakeholders' perceptions regarding the intervention. The solicitation of stakeholder feedback revealed more intentional responses and improved system usefulness. The unit staff reported decreased smartphone noise and environmental noise by shifting the alerts from all caregivers for every patient.

Return on Investment

Return on investment and the value proposition reflect both tangible and intangible costs. Tangible costs are easily measured, whereas intangible costs are not. Addressing alarm management, specifically bed exit policies and workflows, can impact patients, clinicians, financial spending, and regulatory compliance. Savings can be shifted to advancing the identification and development of measures that lead to safer HIT-enabled systems.²¹ Bed exit alarms are used to prevent patient falls. Scholars have estimated that a fall with injury can cost the hospital approximately $14 000 per event.¹⁰ By reducing nuisance alarms, there is an opportunity for more meaningful signals that create better awareness of patient deterioration or adverse events. As identified in domain 2 of the HITS framework, addressing usability and mitigating barriers to use can improve satisfaction for end-users.²¹ False alarms also distract nurses and prevent them from delivering more important patient care.¹²

DISCUSSION

To summarize the results of the effect of the interventions, the analysis revealed improvements in terms of meaningful reduction of nuisance alerts and not increasing or introducing new risks. The informaticist identified components of a bed exit wireless signal pathway that required standardization for policies and workflows, provided staff education on device practices and alarm management expectations, and tailored smartphone alerts to make them more relevant for staff.

The main finding of the current study is based on the premise that holistic changes are needed to reduce nuisance alarms. Consideration must be given to the interdependent relationship of the different technological and human factors in healthcare. Based on the work of Singh and Sittig,²¹ complex adaptive systems are a “collection of individual agents with freedom to act in ways that are not always predictable, and whose actions are interconnected so that one agent's actions change the context for other agents.” The deployment of large-scale technology systems such as call lights and integrated smartphones influences the reach of meaningful communication and collaboration but can also increase risk. Scholars have highlighted in the literature that similar process improvement methodologies can be used across multiple devices to address alarm fatigue that occurs with these HITs.^26–28 The findings of the current study support the interdependent nature of effective alarm management, which is consistent with the premise of the sociotechnical model.

Finally, the findings revealed opportunities concerning continuous improvement of their system and workflow pertinent to alarm management. This single initiative has shown a problem and that a measurable impact can be made using data and continuous performance improvement. This finding is directly related to the third domain of the HITS framework, which highlights the importance of measuring and monitoring technology to ensure continued safety. For the current study, this domain applies to intended objectives such as patient-facing measures and other data points such as system performance and process measures related to poorly defined workflow or policy. It is necessary to mature data analytics to create a safer environment for clinicians and patients. This includes the socialization of current work within the organization and vendor partners to demonstrate the problem, approach, and success to gain support for more comprehensive data analytics and operational resources.

Limitations

There are several limitations related to this nonexperimental, mixed-methods study in practice. First, the sociotechnical models and frameworks were assessed in adult inpatient units, thereby limiting the generalizability of findings and interventions to other clinical settings. Second, the quantitative phase of the current study was nonexperimental in nature. This means that extraneous variables were not controlled. Despite the lack of rigor associated with nonexperimental research, more rigorous experimental studies with control groups can be initiated about this topic due to the findings generated in this study. Third, workflows and wireless configurations are based on current product features; thus, they may not be possible with other technologies or future upgrades. Fourth, a formal survey was not developed for preimplementation and postimplementation to assess nursing perception of change for this specific call type. Lastly, this project required multiple sources of data from multiple systems to ensure a comprehensive assessment. This inconvenience exists at the hospital and enterprise levels, making it even more difficult to benchmark and generalize signal counts for complex environments such as nursing units.

CONCLUSIONS

The results of the present study led to the identification of the components of a bed exit wireless signal pathway that required standardization for policies and workflows. The intervention implemented in this study entailed providing staff education on device practices and alarm management expectations, as well as tailoring the smartphone alerts to make them more relevant for the staff. This quality improvement effort was effective based on the significant reduction in nuisance alerts, positive feedback from staff, and a lack of evidence indicating the introduction of new risks or the increase in existing risks.²⁹ Organizational learnings have been shared with vendor partners, including executive audiences, on current installs and strategic data and interoperability needs that scale-safe and mature alarm management systems. There is demonstrated value in documenting the approach to using sociotechnical models across new and various technologies to support organizational learnings as an industry, support policy at a higher level (e.g., government interoperability standards), and drive vendor accountability for incorporating user needs into development and measurement.^30,31

Based on the findings reported in this study, the informaticist has devised several recommendations for future evaluation areas. First, as broader research is published for other devices and alarms, shared learnings can enhance sociotechnical assessments and guidelines that have primarily focused on EHRs.²¹ Second, future evaluations can explore how interdisciplinary communication and collaboration among staff can affect the management of alarms in healthcare. Investigators can examine the role of effective communication among nurses, physicians, and other healthcare professionals in optimally using alarms.

To conclude, it is not reasonable to assume the nurse chooses to ignore or not manage the alarms appropriately until technology components have been ruled out. By assessing why and how someone was told or expected to do something initially (ie, policy) and then observing their process (ie, workflow), factors in the other dimensions may be revealed.