Attitudes and Practices Related to Clinical Alarms

In 2005, the HTF Clinical Alarms Committee developed an online survey for hospital personnel that addressed clinical alarms.¹²  We administered it by using SurveyMonkey from August 15, 2005, to January 15, 2006, and then again from August 8, 2011, to September 10, 2011. The initial section elicited information on work-related demographics of respondents and their workplace. The next section provided a number of general statements about clinical alarms and prompted respondents to rate their level of agreement with the statement, using options of strongly agree, agree, neutral, disagree, and strongly disagree. The last section contained a list of 9 issues that potentially inhibit effective management of clinical alarms. We asked respondents to rank the issues on a scale of 1 (most important) to 9 (least important). The survey also contained spaces for respondents to provide comments. The 2011 survey repeated most of the questions from the 2005–2006 survey¹²  so that we could track changes. In addition, we added 4 new questions to explore issues that arose from the initial survey:

1. Has your institution experienced adverse patient events in the last 2 years related to clinical alarm problems?

2. Does your institution use “monitor watchers” in a central viewing area to observe and communicate alarm conditions to caregivers?

3. Has your institution developed alarm improvement initiatives over the past few years?

4. Has your health care institution instituted new technological solutions to improve clinical alarm safety?

We also expanded the job classifications and departmental choices on the basis of the first survey.

A number of health care organizations (Table 1), including the American Association of Critical-Care Nurses, supported the study by making members, subscribers, and other stakeholders aware of the survey. A link to the survey was posted on the HTF website and on the websites of these other organizations. We made paper copies available for people who were unable to use the electronic version. The survey elicited no personally identifiable information or the identity of the respondent’s institution. The study was considered exempt by the human investigations committee at Yale University.

We downloaded the data from SurveyMonkey to Microsoft Office Excel 2007 and then to SAS 9.1 for analysis. We used χ² analysis, with odds ratios and 95% confidence intervals as appropriate, to determine differences between the 2005–2006 and 2011 surveys for nominal level responses. For opinions about alarms, where the responses were on a 5-point Likert scale from strongly agree to strongly disagree, we used χ² analysis to compare percentages responding to each level and Mann Whitney U with median interquartile ranges to compare median values.

We received 1327 responses to the 2005–2006 survey and 4278 responses to the 2011 survey. Demographic characteristics of the entire sample are displayed in Table 2. Overall, a vast majority of respondents worked in acute care hospitals and about half worked in intensive care units (ICUs). About 80% of respondents were registered nurses or respiratory therapists. Almost three-quarters had more than 11 years of experience.

A comparison of these characteristics between 2005–2006 and 2011 is shown in Table 3. Compared with the 2005–2006 survey, the 2011 survey revealed that a significantly higher proportion of respondents worked in acute care hospitals and ICUs. Respondents to the 2011 survey were significantly more experienced, with a higher proportion having worked more than 11 years. On the other hand, the 2005–2006 survey had a significantly higher proportion of nurses responding (51.81% of total respondents in 2005–2006 vs 33.06% of total respondents in 2011), although the total number of nurse respondents was greater in the 2011 survey (688 in 2005–2006 vs 1414 in 2011). This smaller proportion of nurses responding to the 2011 survey is most likely due to the large numbers of respiratory therapists (n=2116, 60.75%) who were encouraged to participate by the American Association for Respiratory Care in 2011.

Respondents were asked about their level of agreement with 19 statements about alarms (Table 4). Many of the statements revealed no significant differences between the 2 survey years, although some differences were apparent. Respondents to the 2011 survey were significantly more likely to agree with statements about alarm sounds differentiating the priority of alarm and the helpfulness of central alarm management. Respondents in 2011 were significantly less likely to feel that nuisance alarms occur frequently and disrupt patient care. The situation in 2011 also seemed better regarding the perceived complexity of setting alarm parameters. On the other hand, respondents to the 2005–2006 survey were significantly more likely to agree that that sounds and displays should be distinct based on the parameter or source, policies and procedures regarding alarm management were effectively used, and that there is a requirement to document that alarms are set and appropriate for each patient.

Respondents also ranked the importance of 9 different alarm issues (Table 5). In both 2005–2006 and 2011, they ranked “Frequent false alarms, which lead to reduced attention or response to alarms when they occur” as the most important. The 3 least important issues (“Difficulty in setting alarms properly,” “Noise competition from nonclinical alarms and pages,” and “Lack of training on alarm systems”) were similar across the 2 surveys. Differences occurred in the middle of the rankings. For example, respondents ranked “Overreliance on alarms to call attention to patient problems” as second in importance in 2005–2006, but it dropped to fifth in 2011. The second most important issue in 2011 was “Difficulty in identifying the source of an alarm.”

Four new questions were included in the 2011 survey (see Figure). Almost 1 in 5 respondents (18%) reported experiencing adverse patient events related to alarms in their institutions in the past 2 years. Nearly half of the respondents (47%) stated that monitor watchers/technicians are used in their institutions. Clinical alarm improvement initiatives were developed in 21% of institutions. Only 19% of respondents reported that their institutions instituted technological solutions to improve alarm safety. Technological solutions implemented included integration of monitor and ventilator alarms within the nurse call system, the use of wireless alarm notification devices such as phones and pagers, installation of remote displays, implementation of alarm consolidation systems, and upgrading or purchasing of new equipment.

The hazards associated with clinical alarms have been known for many years; one of the first reports of a sentinel event related to alarms appeared in 1974.¹³  More recently, both the professional and lay literature have called attention to alarms hazards, with a key issue being alarm fatigue. The rationale for the Joint Commission’s new NPSG¹¹  is strong evidence that alarm management is a critical safety issue.

The substantial increase in the number of responses from the first to the second survey was most likely due to effective publication of the survey by using partner society relationships. We also suspect that because of the increased media coverage and ranking on hazard listings, alarm issues are garnering more attention.

Has the situation related to clinical alarms hazards improved since the 2005–2006 survey? It appears not. A key finding in our data is the limited change in the results from the 2005–2006 survey to the 2011 survey. The 2011 results also reveal that only a small percentage of hospitals report alarm improvement initiatives and nearly 20% of the respondents had adverse events related to alarms. The relatively high proportion of respondents reporting adverse events may be due to an increased awareness of alarm safety.

The primary demographic change between the 2 surveys was the participation of respiratory therapists in greater numbers in 2011. A significantly higher percentage of respondents to the 2011 survey (57.56%) were from ICUs, compared with only 31.11% in 2006. In general, ICUs are more affected by the proliferation of alarms because of the density of alarm-generating equipment.

Results of the 2011 survey reveal a significant improvement in the items related to the frequency of nuisance alarms and nuisance alarms disrupting patient care. Possible reasons for this improvement include devices being better able to filter noise and analyze signals, improved preparation of patients for monitoring or therapy, or increased use of monitor watchers to respond to alarms. On the other hand, the responses to the item addressing the impact of nuisance alarms and the possibility that they might lead staff to turn off alarms inappropriately did not change from the first to second survey.

The perceived effectiveness of alarm management policies and procedures decreased in the 2011 survey. The perception of the reduced effectiveness of policies and procedures may be related to an increase in the proportion of patients having devices with alarms. It may also be associated with the competing demands of other safety initiatives, such as initiatives related to the prevention of falls, hospital-acquired infections, and pressure ulcers. It is likely that the emphasis by hospitals on alarm safety, as reflected in the development of effective policies and procedures, may improve with the adoption of the Joint Commission’s NPSG on alarm management.¹¹  The surveys also indicated that institutional requirements to document alarm settings and appropriateness for each patient decreased. This decrease may be due to a perception at institutions that nurses spend more time documenting patient care than providing patient care.

In terms of the survey questions ranking the importance of different alarm issues, frequent false alarms was ranked first in importance in both surveys. This issue clearly must be addressed. Hospitals need to use an interdisciplinary approach to develop solutions to false alarms in their institutions. Researchers should carry out rigorously designed multisite clinical trials with the focus on patients’ outcomes, rather than just on the reduction in the number of alarms. We need to be assured that patients are not harmed by decreases in the number of alarms. Statistical power may be lacking for outcomes such as mortality and sentinel events. We need large multisite studies and/or appropriate surrogates for these relatively rarely occurring outcomes. It will also be important to measure changes in outcomes over time.

In the meantime, bedside clinicians can implement common-sense strategies, such as improving the preparation of patients’ skin before applying monitor electrodes and changing electrodes and telemetry unit batteries daily. They can customize alarm thresholds to the individual patient to avoid nonactionable alarms and consider establishing unit-specific default thresholds. They should also make informed decisions regarding when to initiate and terminate monitoring to avoid unnecessary monitoring with the accompanying false alarms.^14,15  In non-ICU settings, requiring an order to continue monitoring after a certain period of time is one way to address unnecessary monitoring.

A decrease in false alarms should result in a marked reduction in alarm burden with a higher proportion of clinically meaningful alarms. With fewer alarms, nurses’ response time to these clinically meaningful alarms should be shorter. The future impact will most likely be more efficient use of nurses’ time, a quieter environment, and fewer sentinel events related to monitor alarms.

The statements “Difficulty in identifying the source of an alarm” and “Difficulty in hearing alarms when they occur” both moved up in the rankings of importance in 2011. This increase in perceived importance may be related to the increase in the number of monitored patients, the lack of standardized alarm annunciation, and the use of more and more devices with alarms. The increasing levels of background noise in hospitals may also play a role.

Answers to questions unique to the 2011 survey revealed that 671 of 3740 respondents (18%) knew of adverse events related to clinical alarm problems in their institutions in the previous 2 years. Because sentinel events related to alarms are most likely underreported, this statistic may be even more representative of the true extent of the problem than the 566 deaths related to monitor alarms from 2005 to 2008 reported by the FDA.⁸ 

Of marked concern given the prevalence of adverse events related to alarms is the small proportion of hospitals actively working on alarm improvement initiatives. One reason may be the complexity of the issue and the need for a consistent, disciplined, and thorough interdisciplinary approach. Another reason for the lack of alarm improvement initiatives may be the presence of numerous competing priorities. These priorities are likely to change with the need to address the Joint Commission’s 2014 NPSG on alarm management.¹¹  Sources of alarm hazards include sensor artifact, poor human factors design, complexity of ancillary alarm systems, inadequate facilities design, environmental noise, the lack of well communicated or poorly designed care management and alarm escalation processes, and limited staff education on alarms and equipment with alarms.

Individual hospitals need to take a systematic, interdisciplinary approach to alarm safety issues. Focus on alarm events and frequency of alarms, current alarm management practices, staff knowledge, and hospital “culture” is a necessary step. A failure mode and effects analysis might be performed that would lead to strategies and actions suited to individual institutions to reduce alarm hazards.

The use of monitor watchers in a central viewing area to observe and communicate alarm conditions to caregivers was reported by 47% of respondents. Despite the proliferation of hospitals using monitor watchers, it is not known if they improve patients’ outcomes. In studies conducted more than 15 years ago, the presence of monitor watchers resulted in the improved detection of arrhythmias. However, the use of monitor watchers was not associated with lower rates of most adverse outcomes, although fewer episodes of sustained ventricular tachycardia occurred when someone was watching the monitor at all times.^16,17  Research on the use of monitor watchers in current hospital environments needs to be conducted. Is the expense associated with employing monitor watchers a good investment? Does the use of monitor watchers result in improved alarm management and better outcomes for patients?

Several major institutions have implemented long-term comprehensive improvement projects that have demonstrated reductions in the numbers of both false alarms and total alarms, while simultaneously not missing any critically significant true alarms. For example, Johns Hopkins Hospital in Baltimore, Children’s National Medical Center in Washington, DC, Beth Israel Deaconess Medical Center in Boston, and Christiana Care Health System in Delaware embarked on long-term improvement projects that have produced measureable and sustainable results. Despite the reduction in alarms, they report that clinical care has been improved and true alarms have not been missed. Patient and staff satisfaction have both increased, as the noise and frequency of alarms have been reduced.^18–21  Their work has generated strategies that other institutions can implement.

This study has both strengths and limitations. The national survey was broadly distributed to health care professionals working in clinical environments where alarms have proliferated in recent years. Although the 2011 survey had more than 3 times the number in the initial survey, it still represents only a tiny fraction of clinicians in the field.

Clinical alarm safety has been an ongoing challenge that has received considerable attention, but it appears that little measurable progress has been made. Since 2005–2006, when the first survey was conducted, not much has changed except at a few major health care institutions. False and nonactionable alarms continue to contribute to a noisy hospital environment, and sentinel events related to alarm fatigue persist. Perhaps hospitals will pay more attention to this significant patient safety issue with the implementation of the Joint Commission’s NPSG on alarm management.¹¹  Our 2 surveys substantiated the fact that alarm hazards are a problem. We need to act now to attain the goal established by Mary Logan, president of the Association for the Advancement of Medical Instrumentation, at the 2011 Alarm Summit: by 2017, no patient will be harmed by adverse alarm events.²² 

The authors are grateful for the statistical consultation provided by Kristopher Fennie, PhD, Florida International University, Miami, Florida.