Bag Valve Mask Ventilation as a Perceptual-Cognitive Skill

Abstract

Objective

This study used a high-fidelity infant mannequin to examine the relationship between the quality of bag valve mask ventilation (BVMV) and how providers of varying levels of experience use visual feedback (e.g., electronic vital signs) to guide their performance.

Background

BVMV is a common and critical procedure for managing pediatric respiratory emergencies. However, providers do not consistently deliver effective BVMV. Efforts to improve BVMV have ignored the question of how providers effectively use feedback often available during BVMV.

Method

Six expert and six novice respiratory therapists completed two simulations of an infant requiring BVMV. In one, the technology failed to display SpO₂, an important but somewhat redundant visual cue. Eye movements, verbal reports, and ventilation rate (in breaths per minute) were measured in each simulation.

Results

Regardless of SpO₂ availability, eye movements and verbal reports suggested that novices depended strongly on electronic vital signs and when SpO₂ was absent ventilated at a faster rate (exceeding the recommended range of ventilation rates) than when SpO₂ was present. Experts’ ventilation rates were comparable and within the recommended range in both conditions. When SpO₂ was absent, experts emphasized information from direct observation of the patient that novices neglected.

Conclusion

Individual differences in the use of feedback during BVMV contribute to the quality of BVMV. This work bears on the theoretical discussions involving the use of automation and nontechnological cues to guide performance.

Application

These results have the potential to expand the current understanding of factors underlying effective BVMV with implications for training novice providers.

Keywords

expert-novice differences human-automation interaction critical care emergency medicine and resuscitation medical simulation/training and assessment pediatrics and neonatology skilled performance eye tracking

The effective manual ventilation of infants experiencing respiratory failure is an essential, preemptive action for preventing deterioration to cardiopulmonary arrest (Davidovic, LaCovey, & Pitetti, 2005). In the initial management of such emergencies, a bag valve mask (BVM) is the device most commonly used to provide manual ventilation (Davidovic et al., 2005; King & Reynolds, 2008). A BVM consists of three standard components: a self-inflating bag, a face mask, and a valve connecting these two components. Once the face mask is positioned properly on a patient, the provider then breathes for the patient by compressing the bag at a prescribed rate and with enough force to observe the patient’s chest rise.

However, there is considerable evidence that a variety of health care professionals do not consistently provide adequate BVM ventilation (BVMV), delivering substandard ventilation rates or volumes (Davidovic et al., 2005; Elling & Politis, 1983; Harrison, Maull, Keenan, & Boyan, 1982; Niebauer, White, Zinkan, Youngblood, & Tofil, 2011). Ineffective BVMV may not only result in a failure to achieve cardiopulmonary stability but also complicate the patient’s condition. For example, hyperventilating a patient can cause gastric distension, which increases the risk of regurgitation, and increased intrathoracic pressure, which reduces cardiac output and consequently, perfusion of blood to the organs (Kleinman et al., 2010).

To remedy this issue, several studies have attempted to align the ventilation rates or volumes delivered by providers with recommended guidelines using various feedback mechanisms during BVMV. The types of feedback mechanisms have included adding a whistle attachment to indicate the exhalation phase of breathing (Lampotang, Lizdas, Gravenstein, & Robicsek, 2006), capnography monitoring (Hawkes et al., 2013), displaying respiratory function parameters (Kattwinkel, Stewart, Walsh, Gurka, & Paget-Brown, 2009; Schmölzer et al., 2012), and a metronome indicating the proper ventilation rate (Cocucci et al., 2015). Although each of these studies demonstrated an improvement in performance when feedback mechanisms were present, efforts to improve BVMV via feedback have focused on adding mechanisms and ignored the question of how providers effectively make use of the information that is often already available to them during BVMV.

That is, no research to date has appreciated BVMV as a perceptual-cognitive skill and examined it as such. Visual cues that are often available to providers during BVMV (particularly in-hospital providers) come from the patient, such as chest rise and coloration (i.e., cyanosis), and an electronic vital signs monitor, such as pulse oximetry (SpO₂), blood pressure, and heart rate (King & Reynolds, 2008). By recognizing patterns of cues that are indicative of inadequate BVMV, providers can make critical adjustments to their technique, such as reassessing the tightness of the facemask seal or the rate and volume of their ventilations. For example, hyperventilation can decrease cardiac output, which manifests as changes in heart rate, electrocardiogram waveform, and blood pressure (Aufderheide & Lurie, 2004).

Recognizing BVMV as a perceptual-cognitive skill brings the literature on skilled performance in perceptual-cognitively demanding domains, such as transportation and sports, to bear on the question of how to improve the delivery of BVMV in practice. Specifically, experts have been found to be more efficient than nonexperts at extracting relevant visual cues and using them to maintain their performance (transportation: Crundall et al., 2012; Durso & Dattel, 2006; sports: Mann, Williams, Ward, & Janelle, 2007; Williams, Ford, Eccles, & Ward, 2011), particularly when the demands of a situation change. In the domain of transportation, for example, experts have been shown to modulate their gaze behaviors as a function of task complexity (e.g., changing sea or road conditions), whereas the gaze behaviors of novices were largely insensitive to such changes (Godwin et al., 2013; Underwood, 2007; Yang, Kennedy, Sullivan, & Fricker, 2013). The inflexibility of novices may be attributable to their underdeveloped mental models, in other words, their long-term knowledge structures that drive attention to situationally relevant information (Durso & Dattel, 2006; Underwood, 2007). The better developed mental models of experts, however, afford them greater flexibility with the information they can use, making their performance more resilient than that of novices (Underwood, 2007; Williams et al., 2011).

In this vein, the present study examined the relationship between the performance of BVMV and visual cue usage in expert and novice providers. We reasoned that if the performance of BVMV depends on visual feedback, then perturbing the feedback available to a provider should impair the quality of BVMV if the provider cannot effectively make use of the remaining visual cues (i.e., adapt his or her cue sampling strategy). To this end, the availability of pulse oximetry (i.e., SpO₂) was manipulated as an equipment malfunction in two simulations of an infant in respiratory failure.

SpO₂ is a measure of how effectively oxygen is being transported in a patient and has been recognized as an important cue to monitor during BVMV (King & Reynolds, 2008). However, SpO₂ is not (and should not be) the only evidence for the effectiveness of manual ventilation as it is a lag indicator of oxygenation. The rise and fall of the chest is particularly important to monitor as it provides immediate feedback, but other cues, such as patient coloration (i.e., cyanosis) and electronic vital signs (e.g., heart rate), also provide valuable information about the patient’s state (King & Reynolds, 2008).

Two complementary process-tracing measures were used to quantify the use of visual cues and characterize cue sampling strategies—eye tracking and verbal report. Eye tracking provides a measure of how visual attention is allocated in a scene or display, whereas verbal report reveals the specific information that subjects indicate using when performing a task (i.e., the information presumably guiding visual attention; Williams & Ericsson, 2005).

Based on the expertise literature, we predicted that experts would be no worse than novices in either condition and likely better in both. However, the disparity in performance between experts and novices should be greater when SpO₂ is absent from a simulation than when SpO₂ is present. We also expected experts to be more variable in their cue sampling strategy between conditions, suggesting greater adaptability to task demands than novices, whose strategy should be relatively more rigid between conditions. This lack of flexibility is expected to underlie the anticipated greater disparity in performance between experts and novices when SpO₂ is absent. Demonstrating a difference in the ability of experts and novice providers to adapt to task demands would extend similar findings from other domains (Godwin et al., 2013; Underwood, 2007; Yang et al., 2013) to health care and moreover, provide a novel training objective during BVMV instruction (e.g., to learn resilient cue sampling strategies).

Method

Subjects

Subjects were a volunteer sample of 12 respiratory therapists, 6 experts and 6 novices, who were employed at Navicent Health in Macon, Georgia. Experts (M = 12 years of professional experience, SD = 2.77) were required to have at least 8 years of professional experience working in a critical care environment, whereas novices (M = 0.57 years, SD = 0.45) were required to have no more than 2 years of professional experience working in a critical care environment. The average age of experts was 40.67 years (SD = 8.58), and the average age of novices was 26.17 years (SD = 4.30). All subjects reported normal or corrected-to-normal vision and were actively certified in Basic Life Support. Our sample consisted exclusively of respiratory therapists, the health care providers who frequently perform the clinical assessment and initial airway management of acutely ill patients (Kleinman et al., 2010). Study procedures were approved by the Institutional Review Board at the Georgia Institute of Technology and Navicent Health. Informed consent was obtained from each subject. All subjects received monetary compensation for their participation.

Apparatus

Simulations were conducted using a high-fidelity infant mannequin (SimBaby, Laerdal), which was connected to a bedside monitor that displayed simulated electronic vital signs. The simulations occurred in a room that closely resembled an actual patient room, replete with standard medical equipment. Eye movements during the simulations were recorded with a head-mounted mobile eye tracker (Mobile Eye XG, Applied Science Laboratories), which afforded subjects complete freedom of movement. The mobile eye tracker interfaced with a portable CPU, which was worn in a carrying case secured at the subject’s waist.

Procedure

All subjects completed both conditions in a counterbalanced order. To help equate prior knowledge about the simulator, each subject was given a scripted overview of the simulator, an opportunity to ask questions, and a quiz on the simulator’s capabilities. All subjects attained a perfect score on their first try. Following orientation to the simulator, subjects were fitted with the mobile eye tracker. The experimenter then calibrated the eye tracker to the subject’s point of gaze by having the subject fixate on a predetermined sequence of points on the monitor and the mannequin.

Next, subjects completed two simulated scenarios in which an infant experiencing respiratory failure required effective BVMV to be stabilized. By embedding the task of BVMV in two clinical scenarios, actual conditions in which this procedure is performed were approximated, thereby increasing the ecological validity of the experiment. Subjects were blind to the purpose of the study and instructed to assess and treat the patient just as they would if it were an actual patient. A confederate nurse from the Pediatric Intensive Care Unit was present during both scenarios to perform a scripted patient handover to the subject at the beginning of each simulation (see Appendix). Thereafter, the nurse could only repeat information from the patient’s history and give equipment to the subject if asked to do so.

The two scenarios were written and programmed by an attending pediatric critical care physician at Navicent Health. The cause of respiratory failure in the first scenario was due to parenchymal lung disease from pneumonia. The vital signs monitor indicated that the patient’s heart rate was 155 beats per minute with a sinus rhythm, SpO₂ of 80%, blood pressure of 90/55 mmHg, and respiratory rate of 12 breaths per minute. In the second scenario, the cause of respiratory failure was hypoventilation from a sedative overdose. The vital signs monitor indicated that the patient’s heart rate was 110 beats per minute with a sinus rhythm, SpO₂ of 80%, blood pressure of 66/30 mmHg, and respiratory rate of 6 breaths per minute. In both scenarios, the patient’s condition was programmed to deteriorate further if no action was taken in the 2 minutes following the patient handover.

In both scenarios, the infant mannequin displayed two cues—cyanosis, which was simulated by blue LED lights inside the mouth, and chest rise. An electronic vital signs monitor was located at the foot of the patient’s bed and displayed three cues—heart rate, noninvasive blood pressure, and respiratory rate (see Figure 1). In addition, the availability of SpO₂ on the monitor varied between conditions. When SpO₂ was unavailable, the numeric value and waveform were replaced by a question mark, consistent with a nonfunctioning pulse oximeter. Thus, either five or six cues were available in each simulation.

Figure 1.

Configuration of the vital signs monitor in the simulations (as seen by subjects) displaying (upper left) heart rate, (upper right) pulse oximetry, (lower left) noninvasive blood pressure, and (lower right) respiratory rate.

Simulations were controlled by an experienced pediatric critical care nurse who was frequently involved in simulation-based training at the study site. When BVMV began, the operator continuously assessed the quality of manual ventilation based on the displayed respiratory rate and volume. SimBaby detects when an adequate amount of air pressure has been pushed through its airway and displays this information on the operator’s monitor, which was recorded during each simulation for later playback. The operator followed the guideline that infants are to be ventilated at a rate of 20 to 30 breaths per minute during BVMV (Abramo & Cowan, 2002), which is endorsed by the American College of Emergency Physicians. The operator activated a preprogrammed “recovery trend” upon observing five consecutive breaths on the monitor delivered at an appropriate rate and volume. If at any time the subject hyper- or hypoventilated the mannequin, however, the operator activated a trend in which the vital signs (e.g., SpO₂, heart rate, and blood pressure) mimicked the appropriate physiological responses. Failure to properly ventilate the mannequin (i.e., SpO₂ dropping below 50% at any time) ended the simulation. Otherwise, the simulation concluded once the vital signs returned to their normal values.

Immediately after each simulation, subjects were moved to a separate room to watch a video recording of their simulation with their point of gaze superimposed continuously on the scene. Prior to watching the video, subjects were instructed to “think aloud” about what they were doing and thinking during the simulation. Their audio commentary was recorded by the computer.

Analyses

Ventilation rate

Ventilation rate in breaths per minute (bpm) was the measure of the quality of BVMV. To determine the ventilation rate, two researchers naïve to the purpose of the study independently reviewed recordings of the simulation operator’s computer screen from each simulation for each subject in both conditions. Ventilation rate was calculated by counting the number of breaths occurring between the first delivered breath until the moment the recovery trend was activated by the simulation operator, dividing this number by the duration of that window of time, and scaling it in breaths per minute. The single-measures intraclass correlation (ICC) between the two raters’ counts was .89, 95% CI [.750, .953], corresponding to excellent agreement (Cicchetti, 1994).

Gaze behavior

Video recordings of the subject’s point of gaze in the simulations were used to determine the amount of time spent fixating on the monitor or the patient. If a subject looked elsewhere (e.g., the nurse), that fixation was coded as “Other” and was not included in the analysis of gaze behavior.

Verbal protocol

Think-aloud protocols were coded by two researchers, who independently reviewed video recordings of the protocols and counted the number of times a cue was noted by each subject during each experimental condition. Cues consisted of four electronic vital signs (SpO₂, heart rate, respiratory rate, and blood pressure) and two patient cues (chest rise and cyanosis). Prior to counting cues, a coding scheme was established for categorizing utterances as cues based on the various ways that subjects referred to cues in their verbal protocols. For example, “Checking heart rate” and “Heart rate rising” would both count as noting heart rate, whereas “Checking saturations” and “Sats [sic] not picking up” would both count as noting SpO₂. Hotelling’s T² test for independent samples was used to compare the frequency with which each cue was noted by experts and novices in both conditions. Statistical analyses were performed with SPSS Statistics 22. All statistical tests were two-sided; a P value < .05 was considered statistically significant. Effect size measures partial eta-square (η_p²) and Cohen’s d are reported for ANOVAs and t tests, respectively.

Results

Ventilation Rate

Ventilation rates (in breaths per minute) were analyzed in an Experience (expert, novice) × SpO₂ Availability (SpO₂ present, SpO₂ absent) mixed ANOVA, which revealed a significant main effect of SpO₂ availability, F(1, 10) = 8.71, p = .01, η_p² = .47; a nonsignificant effect of experience, F(1, 10) = .08, p = .78, η_p² = .01; and a significant interaction between SpO₂ availability and experience, F(1, 10) = 11.34, p = .01, η_p² = .53 (Figure 2). The interaction suggests that novices delivered a significantly higher ventilation rate when SpO₂ was absent (M = 34.4 bpm, SE = 3.9) than when it was present (M = 20.7 bpm, SE = 3.2), t(5) = 3.39, p = .02, d = 1.38, whereas the ventilation rates of experts were virtually identical between conditions (SpO₂ present: M = 26.9 bpm, SE = 2.6; SpO₂ absent: M = 26.0 bpm, SE = 2.2), t(5) = .58, p = .59, d = .24. Thus, experts ventilated the patient comfortably within the recommended range of 20 to 30 bpm (Abramo & Cowan, 2002) regardless of the availability of SpO₂. Novices, however, tended to ventilate the patient more than 30 bpm when SpO₂ was absent but ventilated the patient just within range when SpO₂ was present.

Figure 2.

Boxplots of ventilation rates of experts and novices as a function of SpO₂ availability. Dashed gray lines show the recommended range of ventilation rates for infant bag valve mask ventilation (Abramo & Cowan, 2002).

Indeed, in the present condition, experts (M = 16.3 seconds, SE = 2.2) did not trigger the recovery trend any sooner than novices (M = 26.3 seconds, SE = 7.2), t(10) = 1.32, p = .22, d = .76. In the absent condition, however, experts (M = 16.5 seconds, SE = 3.2) were significantly faster than novices (M = 37.8 seconds, SE = 8.1), t(10) = 2.45, p = .03, d = 1.41. Thus, there appears to be agreement between the simulator operator’s judgments and the objective, unbiased measure of BVMV quality.

Gaze Behavior

Gaze behavior was measured in terms of the percentage of dwell time on the electronic vital signs monitor, which was calculated over the duration of time between the first delivered breath until the initiation of the recovery trend. Percentage dwell time was analyzed in an Experience (expert, novice) × Trial (first, second) × SpO₂ Availability (SpO₂ present, SpO₂ absent) ANOVA. Trial was included as a factor to test for carryover effects. A significant Experience × Trial × SpO₂ Availability interaction was found, F(1, 16) = 5.58, p = .03, η_p² = .30.

The three-way interaction was assessed with separate Experience (expert, novice) × Trial (first, second) ANOVAs for the SpO₂ present and absent conditions. In short, these analyses suggest that experts generally looked more at the patient than novices except after carrying over from just having dealt with a malfunctioning monitor, in which case, experts looked at the monitor just as much as novices (Table 1).

Table 1:

Percentage Dwell Time on Monitor as a Function of Experience, Trial, and SpO₂ Availability

	SpO₂ Availability
	SpO₂ Present		SpO₂ Absent
	Trial		Trial
	First	Second	First	Second
	M (SE)	M (SE)	M (SE)	M (SE)
Expert	37 (.07)	96 (.02)	63 (.10)	34 (.05)
Novice	69 (.01)	89 (.10)	76 (.09)	66 (.16)

When SpO₂ was present, there was a marginally significant effect of experience, F(1, 8) = 4.23, p = .07, η_p² = .35; a significant effect of trial, F(1, 8) = 40.82, p < .01, η_p² = .84; and a significant interaction, F(1, 8) = 10.29, p = .01, η_p² = .56, with experts (M = 37%, SE = .07) fixating significantly less on the monitor than novices (M = 69%, SE = .01), t(4) = 4.74, p = .01, d = 3.87, except when the monitor had malfunctioned in the previous trial (experts: M = 96%, SE = .02; novices: M = 89%, SE = .10), t(4) = 1.92, p = .13, d = .57. When SpO₂ was absent, there was a marginally significant effect of experience, F(1, 8) = 4.51, p = .07, η_p² = .36, with experts (M = 48%, SE = .08) fixating less on the monitor than novices (M = 71%, SE = .08); a nonsignificant main effect of trial, F(1, 8) = 3.35, p = .10, η_p² = .30; and a nonsignificant interaction, F(1, 8) = .82, p = .39, η_p² = .09.

Despite the variability in the gaze behavior of experts, these providers, unlike novices, were resilient in their performance across conditions. That experts had a greater tendency to fixate more on the patient than novices provides a clue as to why, but to better understand how this affected their performance, we turn now to the verbal reports.

Verbal Protocols

Table 2 shows the average frequency with which experts and novices noted each cue in both experimental conditions. The percentage agreement between raters coding the verbal protocols was 82%. No significant carryover effects were observed with these data.

Table 2:

Average Frequencies at Which Each Cue Was Noted by Experts and Novices When SpO₂ Was Present and Absent

Cue Type	Cue	SpO₂ Present		SpO₂ Absent
Cue Type	Cue	Expert	Novice	Expert	Novice
		M (SD)	M (SD)	M (SD)	M (SD)
Electronic vital signs	SpO₂	1.50 (1.38)	1.50 (.55)	1.50 (1.38)	0.83 (.98)
	Heart rate	0.17 (.41)	0.67 (.82)	0.50 ** (.55)	1.33 ** (.52)
	Respiratory rate	0.50 (.55)	1.00 (1.26)	0.17 * (.41)	1.00 * (.89)
	Blood pressure	0.50 (.84)	0.17 (.41)	0.33 (.52)	0.33 (.52)
Patient	Chest rise	1.00 (.63)	0.50 (.55)	1.17 ** (.75)	0.33 ** (.52)
	Cyanosis	0 (0)	0 (0)	0.33 (.52)	0.50 (.84)

p < .10. **p < .05.

In the SpO₂ present condition, experts did not differ significantly from novices, Hotelling’s T² = 3.82, F(5, 6) = .46, p = .80. In the absent condition, however, experts differed significantly from novices, Hotelling’s T² = 102.7, F(5, 6) = 8.56, p = .02. Post hoc independent samples t tests revealed that novices noted heart rate significantly more often than experts, t(10) = 2.71, p = .02, d = 1.56, and respiratory rate marginally so, t(10) = 2.08, p = .07, d = 1.20. Experts, on the other hand, noted chest rise significantly more often than novices, t(10) = 2.24, p = .05, d = 1.30.

In the absence of SpO₂, experts and novices noted SpO₂ with similar frequency, t(10) = .96, p = .36, d = .56, suggesting that both groups seemed distracted by the failed technology. Despite this, experts may have maintained their performance in large part by emphasizing a cue directly from the patient that novices neglected—chest rise. Experts noted this cue more often than novices in both conditions, but when SpO₂ was absent, experts noted chest rise significantly more often than novices.

Why chest rise? Chest rise provides immediate feedback on the adequacy of one’s technique; the provider can instantly see if enough air is being delivered to the patient. Therefore, using this feedback should facilitate adequate ventilation. Novices, on the other hand, remained fixated on the electronic vital signs, particularly heart and respiratory rate.

Discussion

The removal of SpO₂ during BVMV evoked different strategies for managing the situation from expert and novice providers. Experts relied on patient observation as their primary source of information and not on the automation. Novices, however, showed a strong and consistent bias toward the automated technology even though evidence was available from patient observation (Mosier, Skitka, Heers, & Burdick, 1998). Although chest rise is an immediate indicator of the quality of BVMV, novices relied on technological feedback, regardless if it was completely functional. In that sense, novices were less flexible than experts as they could not adapt their cue sampling strategy (i.e., they relied principally on electronic vital signs) as effectively as experts, who adapted by placing greater emphasis on salient information from the patient. The relative inflexibility of novices has been observed in other domains such as transportation wherein experts, but not novices, adapt their gaze behaviors as a function of task complexity to maintain performance (Godwin et al., 2013; Underwood, 2007; Yang et al., 2013).

That the performance of novices suffered when SpO₂ was unavailable suggests that there may be a gap in cue sampling skills between novice and expert providers of BVMV. Not only were novices unable to effectively make use of the remaining feedback on the monitor to maintain the quality of their BVMV, their focus on the remaining vital signs may have also come at the expense of noting important cues from the patient, such as chest rise. Of course, this has clinical significance because excessive ventilations during BVMV can negatively impact a patient, for example, by increasing the risk of stomach inflation and consequently, regurgitation and aspiration of stomach contents (Abramo & Cowan, 2002; Kleinman et al., 2010).

The results of the present research should be interpreted in light of its limitations. First, the reliability of our findings should be assessed in a larger sample. However, it is promising that our results were consistent with theoretically motivated predictions and these effects were large enough to be detected with the sample at hand. A second limitation is that a simulator operator determined the quality of BVMV based on objective physiological parameters from the mannequin and controlled the mannequin’s response accordingly. The operator was blind to the expertise of most (but not all) subjects and was not blind to experimental condition. However, determination of ventilation rates by naïve raters showed that the operator’s judgments did indeed correspond to an objective and unbiased indicator of the quality of BVMV. Lastly, our results may not generalize to other manual ventilation devices (e.g., those without a self-inflating bag). For example, the Neopuff (Fisher & Paykel), designed specifically for infant resuscitation, requires the provider to intermittently occlude a T-piece to allow inspiration and expiration.

Understanding the cause(s) of the novices’ automation bias would require further investigation, but a likely explanation is that novices were simply doing what was cognitively easiest for them (Mosier et al., 1998). Determining the adequacy of chest rise, which is subtle in infants, is difficult and subjective (Davidovic et al., 2005). In contrast, electronic vital signs are objective, reliable, and readily perceivable indices of a patient’s status. Therefore, it would not be surprising if novices were allocating their attention to cues that were both informative and had little perceptual uncertainty. This automation bias has also been shown to negatively affect commercial pilots who ignore “obvious” cues in the environment, deciding instead to tunnel on the automation (Mosier et al., 1998). In contrast to our results, Mosier et al. (1998) found that reliance on automated cues actually increased with experience. Our experts, however, were not unduly influenced by automated feedback. Rather, their performance seemed to exemplify the medical adage, “Treat the patient, not the monitor.”

Considering that the novices had roughly half a year’s worth of professional experience, on average, their performance may largely reflect their formal instruction. Indeed, the training of BVMV is often basic, failing to reproduce the complexity and difficulty of the situations in which it occurs (Pastis, Doelken, Vanderbilt, Walker, & Schaefer, 2013). The results of the current study may suggest a deficiency in the current training of BVMV and moreover, offer a measurable behavior (i.e., cue sampling) that should be targeted during instruction. Additionally, making the interpretation of vital signs easier could afford providers, particularly inexperienced ones, more time to assess the patient (e.g., by visualizing trends or patterns in vital signs over time; Drews, 2008).

Key Points

Individual differences in the use of visual feedback contribute to the quality of bag valve mask ventilation (BVMV).

Experts delivered adequate ventilation regardless of SpO₂ availability, whereas novices tended to ventilate the patient too quickly when SpO₂ was unavailable.

When SpO₂ was unavailable, experts adapted by emphasizing information from direct observation of the patient, whereas novices depended strongly on technological feedback, regardless of whether it was fully functional.

Footnotes

Appendix

Script recited by confederate nurse during patient handover in the first scenario (i.e., parenchymal lung disease from pneumonia):

The patient was admitted last night through the emergency room due to sudden onset of a fever of 103°F with 3 days of cough and runny nose. She had a partial septic work-up on admission and I was told on check-out that she had an abnormal chest x-ray. Antibiotics were given on admission. She was placed on oxygen via nasal cannula at 1 liter per minute due to oxygen saturation of 89% while breathing room air. When mom stepped out of the room to smoke, I noticed that her oxygen saturation started to drop, so I increased her oxygen to 2 liters per minute.

Script recited by confederate nurse during patient handover in the second scenario (i.e., sedative overdose):

The patient was admitted this morning for a scheduled Nissen fundoplication due to multiple episodes of aspiration pneumonia. She was placed on NPO at 12 midnight and woke up crying, looking very irritable. Her nurse asked the post-call resident if she could give the patient a dose of Versed or Morphine to calm her down. The resident said yes and ordered the outgoing nurse to give the patient 1 mg each of Morphine and Versed. The outgoing nurse administered the medications and then left. I noticed that the patient’s oxygen saturation is going down.

Acknowledgements

This research was funded internally. The authors would like to thank the confederate nurses for their involvement in the simulations as well as the subjects for their participation. Portions of these data were presented at the 2016 International Symposium on Human Factors and Ergonomics in Health Care in San Diego, California.

Joel M. Mumma is a graduate student in the engineering psychology PhD program at the Georgia Institute of Technology. He received his MS at the Georgia Institute of Technology in 2017 and his BA in brain and cognitive sciences from the University of Rochester in 2010. His research focuses on the cognitive processes underlying skilled performance in health care providers.

Francis T. Durso is a professor in the engineering psychology program at the Georgia Institute of Technology. He received his BS from Carnegie Mellon University and his PhD in learning/cognitive processes from the State University of New York at Stony Brook in 1980. His current research interests focus on cognitive factors used to manage dynamic situations in domains such as aviation, rail, and health care.

Michelle Dyes is a pediatric critical care nurse at Navicent Health. She received her BSN nursing degree from Georgia College in 1995. She received her Pediatric Critical Care Certification (CCRN-P) from the American Association of Critical Care Nurses.

Rogelio dela Cruz is a pediatric critical care physician at Navicent Health. He received his doctor of medicine at De La Salle University in 1986 and completed specialty training in pediatrics at St. Joseph’s Hospital in 1994 and subspecialty training in pediatric critical care at the University of Florida in 2004. His research interests focus on medical simulation.

Valerie P. Fox is a registered respiratory therapist (RRT) and education coordinator in the respiratory department of Navicent Health. She received an associate of science degree in respiratory therapy from Macon State College in 1999.

Mary Hoey is a nurse researcher at the Center for Disruption and Innovation at Navicent Health. She obtained a master’s degree in advanced practice (education) from the University of Dundee in 2012 and a MBA from the University of Hull in 1998. Her research interest is focused on the “transition to practice” experiences of newly licensed registered nurses.

References

Abramo

T. J.

Cowan

(2002). Respiratory failure. In Strange

G. R.

Ahrens

W. R.

Lelyveld

Schafermeyer

R. W.

(Eds.), Pediatric emergency medicine: A comprehensive study guide (2nd ed., pp. 5–10). New York, NY: McGraw-Hill.

Aufderheide

T. P.

Lurie

K. G.

(2004). Death by hyperventilation: A common and life-threatening problem during cardiopulmonary resuscitation. Critical Care Medicine, 32(9), S345–S351.

Cicchetti

D. V.

(1994). Guidelines, criteria, and rules of thumb for evaluating normed and standardized assessment instruments in psychology. Psychological Assessment, 6(4), 284–290.

Cocucci

Madorno

Aguilar

Acha

Szyld

Musante

(2015). A metronome for pacing manual ventilation in a neonatal resuscitation simulation. Archives of Disease in Childhood-Fetal and Neonatal Edition, 100(1), F47–F49.

Crundall

Chapman

Trawley

Collins

van Loon

Andrews

Underwood

(2012). Some hazards are more attractive than others: Drivers of varying experience respond differently to different types of hazard. Accident Analysis & Prevention, 45, 600–609.

Davidovic

LaCovey

Pitetti

R. D.

(2005). Comparison of 1- versus 2-person bag valve-mask techniques for manikin ventilation of infants and children. Annals of Emergency Medicine, 46(1), 37–42.

Drews

F. A.

(2008). Patient monitors in critical care: Lessons for improvement. In Henriksen

Battles

J. B.

Keyes

M. A.

Grady

M. L.

(Eds.), Advances in patient safety: New directions and alternative approaches (pp. 1–8). Rockville, MD: Agency for Healthcare Research and Quality.

Durso

F. T.

Dattel

A. R.

(2006). Expertise and transportation. In Ericsson

K. A.

Charness

Feltovich

P. J.

Hoffman

R. R.

(Eds.), The Cambridge handbook of expertise and expert performance (pp. 355–372). New York, NY: Cambridge University Press.

Elling

Politis

(1983). An evaluation of emergency medical technicians’ ability to use manual ventilation devices. Annals of Emergency Medicine, 12(12), 765–768.

10.

Godwin

H. J.

Hyde

Taunton

Calver

Blake

J. I. R.

Liversedge

S. P.

(2013). The influence of expertise on maritime driving behavior. Applied Cognitive Psychology, 27, 483–492.

11.

Harrison

R. R.

Maull

K. I.

Keenan

R. L.

Boyan

C. P.

(1982). Mouth-to-mask ventilation: A superior method of rescue breathing. Annals of Emergency Medicine, 11(2), 74–76.

12.

Hawkes

G. A.

O’Connell

B. J.

Livingstone

Hawkes

C. P.

Ryan

C. A.

Dempsey

E. M.

(2013). Efficacy and user preference of two CO2 detectors in an infant mannequin randomized crossover trial. European Journal of Pediatrics, 172(10), 1393–1399.

13.

Kattwinkel

Stewart

Walsh

Gurka

Paget-Brown

(2009). Responding to compliance changes in a lung model during manual ventilation: Perhaps volume, rather than pressure, should be displayed. Pediatrics, 123(3), e465–e470.

14.

King

Reynolds

S. L.

(2008). Bag-valve-mask ventilation. In King

Henretig

F. M.

(Eds.), Textbook of pediatric emergency procedures (pp. 109–125). Philadelphia, PA: Lippincott Williams & Wilkins.

15.

Kleinman

M. E.

Chameides

Schexnayder

S. M.

Samson

R. A.

Hazinski

M. F.

Atkins

D. L.

Zaritsky

A. L

. (2010). Part 14: Pediatric advanced life support: 2010 American Heart Association guidelines for cardiopulmonary resuscitation and emergency cardiovascular care. Circulation, 122, S876–S908.

16.

Lampotang

Lizdas

D. E.

Gravenstein

Robicsek

(2006). An audible indication of exhalation increases delivered tidal volume during bag valve mask ventilation of a patient simulator. Anesthesia & Analgesia, 102(1), 168–171.

17.

Mann

D. T.

Williams

A. M.

Ward

Janelle

C. M.

(2007). Perceptual-cognitive expertise in sport: A meta-analysis. Journal of Sport and Exercise Psychology, 29(4), 457–478.

18.

Mosier

K. L.

Skitka

L. J.

Heers

Burdick

(1998). Automation bias: Decision making and performance in high-tech cockpits. The International Journal of Aviation Psychology, 8(1), 47–63.

19.

Niebauer

J. M.

White

M. L.

Zinkan

J. L.

Youngblood

A. Q.

Tofil

N. M.

(2011). Hyperventilation in pediatric resuscitation: Performance in simulated pediatric medical emergencies. Pediatrics, 128(5), 1195–1200.

20.

Pastis

N. J.

Doelken

Vanderbilt

A. A.

Walker

Schaefer

J. J.

III . (2013). Validation of simulated difficult bag-mask ventilation as a training and evaluation method for first-year internal medicine house staff. Simulation in Healthcare, 8(1), 20–24.

21.

Schmölzer

G. M.

Morley

C. J.

Wong

Dawson

J. A.

Kamlin

C. O. F.

Donath

S. M.

. . . Davis

P. G

. (2012). Respiratory function monitor guidance of mask ventilation in the delivery room: A feasibility study. The Journal of Pediatrics, 160(3), 377–381.

22.

Underwood

(2007). Visual attention and the transition from novice to advanced driver. Ergonomics, 50(8), 1235–1249.

23.

Williams

A. M.

Ericsson

K. A.

(2005). Perceptual-cognitive expertise in sport: Some considerations when applying the expert performance approach. Human Movement Science, 24(3), 283–307.

24.

Williams

A. M.

Ford

P. R.

Eccles

D. W.

Ward

(2011). Perceptual-cognitive expertise in sport and its acquisition: Implications for applied cognitive psychology. Applied Cognitive Psychology, 25(3), 432–442.

25.

Yang

J. H.

Kennedy

Sullivan

Fricker

R. D.

Jr. (2013). Pilot performance: Assessing how scan patterns & navigational assessments vary by flight expertise. Aviation, Space, and Environmental Medicine, 84, 116–124.