**4. Patient monitor alarm design**

been particularly important for critically ill patients who require mechanical ventilatory support. In such applications, the monitor is designed to be exquisitely sensitive to detect even the slightest changes in a patient's oxygenation or ventilation status [38]. As demonstrated in *Clinical Vignette #1* later in the chapter, an alarm may be triggered following the detection of a very small respiratory parameter "excursion," regardless of its clinical significance or magnitude of the observed change in the patient's actual clinical status. In this context, apnea and minute volume warnings are among the most common alarms triggered, with majority of such occurrences deemed clinically irrelevant upon further interrogation [39]. Moreover, many VOA triggers can be attributed to artifactual sources (e.g., patient movement, interruption of blood flow by inflating blood pressure cuff, and even atmospheric pressure variations) [37]. Thus, providers should be educated accordingly to ensure that the above considerations are appropri-

Hemodynamic alarms (HA) monitor a variety of parameters, of which the most common ones include heart rate, systolic/diastolic/mean blood pressure, and various other intravascular pressure measurements via both invasive and noninvasive approaches [37, 40]. Hemodynamic monitoring has become a useful tool for the bedside assessment of patients in a number of clinical scenarios, from routine telemetry applications to advanced intravascular catheter utilization. There is some degree of predictability based on measured parameters, especially when trend determination and volume responsiveness are being considered [41, 42]. Hemodynamic monitors are particularly important in the setting of an unstable (or potentially unstable) patient, similar to the one described in *Clinical Vignette #3* later in the chapter. In such capacity, HAs can help facilitate rapid intervention and prompt correction of emergent issues. Still, HAs are far from perfect, with significant shortcomings in their discriminatory capabilities. More specifically, HAs are unable to identify a patient as "stable" or "unstable," especially when physiologic compensatory processes mask any underlying instability or in the setting of rapid change in hemodynamic status [43]. Thus, when using any particular monitoring modality, there is no substitute for an astute clinician who is able

Bed and chair pressure sensor (BCPS) alarms are utilized across many hospitals and other health-care facilities to help reduce mechanical falls among patients who experience ambulatory or balance difficulties [47, 48]. Falls typically occur as patients attempt to mobilize and/or ambulate without the required assistance of trained health-care staff [49]. Consequently, the use of BCPS alarms serves to alert staff—typically by a pressure-sensitive mechanism—when a patient attempts to move from a bed or chair without assistance. However, the weightsensitive pads are easily triggered by very slight patient movement, resulting in a significant number of false alarms [50, 51]. This challenge was readily apparent in *Clinical Vignette #3* later in the chapter, as the majority of BCPS alerts were likely due to the patient merely shifting slightly in the bed, and not by an actual attempt to independently mobilize and/or ambulate.

ately factored into final clinical determinations and decisions.

to effectively correlate HA findings with the clinical reality [44–46].

**3.2. Hemodynamic alarms**

96 Vignettes in Patient Safety - Volume 4

**3.3. Bed and chair pressure sensors**

Patient monitors are designed to have high sensitivity to predefined changes in various measured parameters, including vital signs, respiratory/ventilator status, and patient movements. However, the major drawback associated with high alarm sensitivity is the poor specificity and inherently disproportionate number of nonactionable (or nonclinical) alarms triggered [22, 53, 54]. Depending on the specific alarm and clinical setting, the estimated in range of "false positives" may be as high as 80–99% of all triggered alarms [8]. Broadly speaking, nonactionable alarms can be categorized as false alarms, nuisance alarms, and technical alarms (**Figure 1**). To elaborate further, false alarms occur in the absence of an actual patient or system trigger and typically result from a measurement artifact [55]. Technical alarms mandate the provider to attend to some operational aspect of the monitoring system, such as when readjustment of monitor leads/sensors is required [21]. Nuisance alarms are defined as clinically insignificant alarms that may interfere with patient care [10]. In aggregate, these nonactionable alarms are a major cause of the overall desensitization of hospital staff that may ultimately result in AF (**Figure 2**).

Furthermore, to be effective, the alarms transmitted by monitoring systems must trigger some degree of cognitive response in health-care providers. This equates to introducing stress and the need for constant vigilance, both of which further heighten the risk of AF [56, 57]. When multiple clinical competing priorities collide, it becomes increasingly difficult for a provider to proactively address all ongoing problems, thus forcing them to resort to only partially addressing acute issues while at the same time disrupting other (parallel) activities due to multitasking [58–61]. Consequently, an ideal alarm should be perfectly audible and easily recognized by health-care providers working within the patient care unit [8], while at the same time minimizing the amount of stress imposed on the responding clinical staff.

The increasingly complex environment of modern health-care systems has led to several important considerations regarding the practical application of monitoring systems. For example, space-related issues deserve special mention, with overly crowded clinical units creating an abundance of alarm-related stimuli and geographically larger clinical units

hospital for evaluation. Upon arriving in the emergency department, short-acting bronchodilators and oral corticosteroids were administered with only mild symptomatic improvement. Given the patient's dyspnea at rest, as well as decreased oxygen saturation of 86%, she was admitted to the pulmonology unit. Supplemental oxygen and intravenous corticosteroids were

Combating Alarm Fatigue: The Quest for More Accurate and Safer Clinical Monitoring Equipment

http://dx.doi.org/10.5772/intechopen.84783

99

At admission, continuous pulse oximetry monitoring was started. The patient's hypoxemia seemed to improve slightly over the next 4 days, with oxygen saturation climbing to 88–90% range. Still, the patient's ventilatory monitor sent alarm signals to the hospital staff several times an hour due to high respiratory rate and episodic oxygen desaturations. Alarm signals were transmitted as either a single low tone (respiratory rate) or a double alarm (desaturations), alternating between low and medium tones. The difference of alarm tone indicated the range in which the patient's oxygen saturation was measured, but the assigned night-shift nurse found the tones to be too difficult to distinguish and would routinely just perform an in-person check of the saturation level upon entering the room. Throughout the first two nights, the same nurse responded to the alarms in a timely fashion, only to find the patient stable and with no signs of acute distress. Assuming that alarms are unlikely to represent any actionable clinical events, the same nurse then began to silence the sounds and began checking on the patient hourly. In the early morning hours of the fourth day, the nurse silenced the alarm once again, intending to assess the patient once the remainder of her rounding routine was completed. When the nurse finally came to the patient's room an hour later, she found the patient unresponsive and cyanotic. A rapid assessment showed an oxygen saturation of 79%. The patient was immediately intubated, transferred to intensive care unit, and

**5.2.** *Clinical Vignette #2***: 65-year-old male transferred to inpatient unit following a** 

A 65-year-old male with a history of osteoarthritis of the right knee and refractory pain underwent preoperative evaluation by an orthopedic surgeon. Given his adequate performance status and lack of comorbidities, the patient was determined to be a suitable candidate for total right knee arthroplasty. The surgical procedure was uneventful, with appropriate antibiotic and venous thrombosis prophylaxis administered perioperatively. Following a brief recovery in the postanesthesia care unit, the patient was transferred to the inpatient floor with expected discharge within 5 days postsurgery. Due to the nature of his surgery and apparent fall risk, the patient's room was fitted with weight-sensitive bed and chair alarms. During the first 3 days, he remained relatively sedated due to the frequent administration of pain medications. However, as the patient began to regain strength, his analgesia regimen was tapered. On day 4, the concurrent increase in patient's movement began to trigger his bed monitor to the point where the on-call nurse was receiving nearly constant alarm notifications. Multiple times, the nurse entered to assess the patient only to find him resting comfortably without apparent attempt to leave his bed. Later that night, after leaving the patient's room, the nurse was unexpectedly assigned to three additional patients due to an unplanned absence of a coworker. As the nurse hurried to assess the new patients, the bed monitor transmitted yet another alarm signal. Annoyed by the repeated negative alarms, the nurse disabled the alerts from the bed monitor,

administered.

mechanical ventilation was initiated.

**total knee arthroplasty**

**Figure 2.** Schematic representation of the classification of alarm types triggered by various patient monitoring systems, including both actionable and nonactionable alerts (source: Ruskin [8]; Gorges [66]; and Tsien [67]).

presenting a barrier to prompt patient access. Elevated acuity and high patient throughput are also important considerations in this context [62].

Furthermore, technological advancements facilitated the development of increasingly sophisticated alarm systems, with novel features designed to decrease the nuisance factor of the alert mechanism while preserving the level of overall clinical vigilance [63, 64]. These are intended to provide a range of alarm tones that allow care providers to easily identify and prioritize alarms, typically as high, medium, or low priority. However, the implementation of such systems (e.g., IEC 60601-1-8 standard) has presented challenges in terms of recognizability of melodic alarm tones. More specifically, nurses found it difficult to accurately identify all of the melodic tones signifying high-priority alarms, in addition to the potential for confusion between certain alarm pairs [65]. An example of such phenomenon is presented in *Clinical Vignette #1* below, where two sets of tones were too difficult for the nurse to readily differentiate, rendering the alarm feature ineffective. Consequently, it is important for systems to have some degree of built-in learnability and flexible discriminative ability, with continued refinement, development, and testing of each clinical alarm, both alone and in tandem with other competing alarms [65]. Without exception, any observed deficits in patient monitor effectiveness and/or safety should prompt an immediate critical evaluation of both technical and clinical aspects of its implementation and function.
