|Home | About | Journals | Submit | Contact Us | Français|
Anesthesiology requires performing visually-oriented procedures while monitoring auditory information about a patient’s vital signs. A concern in operating rooms environments is the amount of competing information and the effects that divided attention have on patient monitoring, such as detecting auditory changes in arterial oxygen saturation via pulse oximetry.
We measured the impact of visual attentional load and auditory background noise on the ability of anesthesia residents to monitor the pulse oximeter auditory display in a laboratory setting. Accuracies and response times were recorded reflecting anesthesiologists’ abilities to detect changes in oxygen saturation across three levels of visual attention in quiet and with noise.
Results show that visual attentional load substantially impacts the ability to detect changes in oxygen saturation levels conveyed by auditory cues signaling 99 and 98% saturation. These effects are compounded by auditory noise, with up to a 17% decline in performance. These deficits are seen in the ability to accurately detect a change in oxygen saturation and in speed of response.
Most anesthesia accidents are initiated by small errors that cascade into serious events. Lack of monitor vigilance and inattention are two of the more commonly cited factors. Reducing such errors is thus a priority for improving patient safety. Specifically, efforts to reduce distractors and lower background noise should be considered during induction and emergence, periods of especially high risk, when anesthesiologists must attend to many tasks and are thus susceptible to error.
Anesthesiology is a discipline of medicine that requires intense vigilance, multi-tasking ability, good aural perception and communication skills, as well as critical decision-making. Anesthesiologists are required to balance a multitude of tasks while working in the operating room, with a primary focus on the health and well-being of the patient. These tasks are carried out in a complex environment that is rich in sensory information, some of it critical to the anesthesiologist’s tasks while other information is either irrelevant and/or distracting. As one example of the challenges in such a setting, previous research has found the audible noise level in the operating room to average 77 dB 1, with episodes that can often eclipse 100 dB 2. Not surprisingly, this level of noise has been shown to have a detrimental impact on anesthesiologists’ ability to perform cognitive tasks 1, 2. In addition to noise, the competing attentional demands of the operating room environment may also detrimentally impact anesthesiologists’ performance. Not surprisingly, operating room performance has been previously linked to changes in attentional load mediated by, among other factors, task complexity 3 and workload 4, 5.
The pulse oximeter is perhaps the most important monitor anesthesiologists use, providing information on arterial oxygen saturation, heart rate, and rhythm. When focusing visual and haptic attention on a surgical case, anesthesiologists are often required to rely on their auditory perception of this monitor to detect changes in these physiologic parameters, requiring the detection of an auditory signal within the previously described background noise and with competing attentional demands. Also, anesthesiologists have pressure from surgical colleagues to keep monitors and alarms to a minimum volume despite the frequent presence of loud background noise and music. Consequently, the volume of the pulse oximeter is often relatively low, presenting the anesthesiologist with a classic signal-in-noise challenge. Surprisingly, there is a paucity of work on examining the perceptual expertise of anesthesiologists at perceiving changes in pulse oximetry pitch, and, perhaps more importantly, how background noise and attentional load impact pulse oximeter monitoring. This lack of study is particularly startling given that inattention and lack of monitor vigilance are commonly cited factors implicated during critical incidents in anesthesia 6–9. Coupled with this, previous research has shown that the majority of anesthesia-related accidents are not the product of a single catastrophic error, but instead are derivative of a number of small errors, such as not detecting a change in oxygen saturation, that cascade into a serious event 10. The current study addresses this gap in the literature by investigating the impact of attentional load and auditory noise, as well as the interaction between these factors, on the ability of a cohort of resident anesthesiologists to detect changes in oxygen saturation with pulse oximetry.
Participants included 33 resident anesthesiologists (19 male, mean age = 30±3 years old) who were paid to participate. All recruitment and experimental procedures were approved by the Vanderbilt University Institutional Review Board(Nashville, TN).
All stimuli throughout the study were presented using MATLAB (MATHWORKS Inc., Natick, MA) software with the Psychophysics Toolbox extensions 11, 12, on a Dell computer(Dell Inc., Round Rock, TX). Visual stimuli consisted of individual letters presented in the central visual field in rapid serial visual presentation at a rate of 10 Hz (100 ms per presentation). Visual stimuli included 25 capital letters (excluding “Y” for purposes of disambiguating vowels and consonants in the attentional tasks) presented in either red or white. Visual stimuli were presented on a Samsung Sync Master 2233RZ 120 Hz monitor (Samsung Group, Seoul, South Korea) 0.45 m in front of the participant (Figure 1A). Letters were presented in Geneva, 96-point font, and were approximately 2.2 × 2.5 cm in size (though width varied slightly between letters).
Auditory stimuli were presented via mounted speaker 0.45 m from the participants and at a 90° angle from the participants’ heads on their right side (Figure 1A). Auditory stimuli consisted of 100 ms, sine-wave gated pure tone beeps at frequencies matching the 99% (648 Hz) and 98% (630 Hz) blood-oxygenation saturation levels on a Philips patient monitor (Model MP70; Koninklijke Philips Electronics N.V., Amsterdam, Netherlands) at a rate of 75 beats per minute. Previous research has shown that a large majority of individuals are able to detect such a change13. To determine the fundamental pitch of these levels, sound wavelengths produced by a Fluke Biomedical Index 2MF SpO Simulator(Fluke Biomedical, Everett, WA)at oxygen saturations ranging from 40–100% were measured with a Hameg 507 oscilloscope(HAMEG Instruments GmbH, Mainhausen, Germany). Empirically-measured sound frequencies were then fitted with an exponential function, which was subsequently used to interpolate frequency values for the appropriate saturation levels. This exponential function,
where Frequency represents the sound frequency and x represents the level of oxygen saturation, fit the measured sound frequencies significantly well (R2 = 0.99; Figure 1B). All auditory beeps were presented as pure tones, at 80 dB SPL, corresponding to the default QRS volume setting of 2 on the Philips patient monitor, and confirmed as a common level of usage with participants. It should be noted here that different models of pulse oximeters use distinct sound pressure levels as well as distinct harmonics to signal changes14. Duration and timing of all visual and auditory stimuli were confirmed using a Hameg 507 oscilloscope with a photovoltaic cell and microphone.
Background noise was included in half of the conditions. Background noise consisted of pre-recorded operating room noise during cardiac surgery without pulse oximetry beeps. This background noise was played with an average dB SPL level of 67 (ranging from 58–86 dB SPL) via a Sony MZ-R700 mini disc recorder(Park Ridge, NJ)through an Optimus SA-155 stereo amplifier and Optimus XTS-3 speakers(Optimus Acoustics, Bloemfontein, South Africa). Background noise included sounds of conversations, movement of operating room personnel, and movement of surgical instruments, but specifically excluded alarms. It should also be noted here that this background noise differs from that in the real operating room in that resident participants were not required to interact with signals in the background noise, such as a request from the surgeon.
Participants completed a total of six tasks in a 3 × 2 design. The first factor, attentional load, was varied through three visual tasks presented in the central visual field that are commonly used in studies of attentional effects15–17, and the second factor, noise level, was varied though the presence or absence of pre-recorded operating room background noise. For each task, participants were seated inside an unlit WhisperRoom™ (Model SE 2000; Whisper Room Inc, Morristown, TN) with their forehead placed against a Headspot (University of Houston College of Optometry, Houston, TX) forehead rest locked in place, with a chinrest and chair height adjusted individually to the forehead rest. Participants were asked to fixate towards a fixation cross at all times, and were monitored by close circuit infrared cameras throughout the experiment to ensure compliance.
For the first visual task, which we will refer to as the low-attentional load condition, the participant was asked only to fixate towards a constant fixation cross. For the second visual task, which we will refer to as the medium-attentional load condition, participants were presented with a series of single letters in rapid serial visual presentation format, 96% presented in white and 4% in red (Figure 1C). Participants were asked to respond as quickly and accurately as possible via button press with their left hand any time that they saw any red letter (target rate = 4%). Specific letters and colors were presented in pseudorandom order. For the third visual task, which we will refer to as the high-attentional load condition, participants were presented with a series of single letters in rapid serial visual presentation format, 80% presented in white and 20% in red (Figure 1D). Participants were asked to respond via button press with their left hand any time that they saw a red vowel (with vowels making up 5 of the 25 letters, target rate = 4%). Specific letters and colors were presented in pseudorandom order. A total of 6000 letter presentations were made for each condition, with 240 targets.
With each of these 6 tasks, individuals were also asked to complete an auditory monitoring task that was identical across conditions. Auditory beeps were presented continually at a rate of 75 beeps per minute, with 90% of the beeps (675) presented at the 99% saturation pitch and 10% at the 98% saturation pitch (75), for a total of 750 beeps during each 10-minute trial. Participants were asked to respond as quickly and accurately as possible via button press with their right hand every time they heard a 98% saturation-level beep. Targets were presented in a pseudorandom order.
Each run began with an instruction screen, after which the participant was asked if he or she understood the instructions. After indicating they were ready by pressing the spacebar, participants were cued with a timed visual countdown, and the monitoring task began. Half of the way through, at five minutes, participants were offered a break, if needed. After indicating readiness to continue, the second half of the given condition proceeded in the same manner as the first half. The order of the six conditions was randomized across participants. Between conditions, participants were offered breaks as needed. Total experimental time lasted 60 minutes not including breaks or instructions.
For each of the conditions, low, medium, and high attentional load with and without noise, mean accuracies and response times (RTs) were calculated for the auditory task. Repeated-measures, 3 × 2 ANOVAs were run testing for main effects of both attentional load and noise, as well as an interaction between these two factors. Where these ANOVAs were significant (α= 0.05), follow-up, pair-wise t-tests were conducted.
In order to assure that the visual attentional load tasks did, in fact, vary in difficulty, mean accuracies and RTs were calculated for the visual task for four of the six conditions, medium and high attentional load with and without noise, (no responses to visual stimuli were made to during the low attentional-load conditions). Repeated-measures, 2 × 2 ANOVAs were run testing for main effects of both attentional load and noise as well as an interaction between these two factors. Where these ANOVAs were significant, follow-up, pair-wise t-tests were conducted.
Mean accuracies of individuals’ ability to detect changes in pitch associated with changing levels of arterial oxygen saturation were calculated for each of six experimental conditions. A 3 × 2 ANOVA showed a significant main effect of visual attentional load (F =11.90, p < 0.01) and of audible noise level (F = 56.51, p < 0.01). This analysis also revealed that these effects were additive, showing no significant interaction (F = 1.29, p = 0.28). Follow-up t-tests showed that participants performed with lower accuracies on the high relative to the low attentional load task, and accuracies were lower in the conditions with noise relative to those without (Figure 2A).
An analysis of mean RTs provided similar findings (Figure 2B). Once again a 3 × 2 ANOVA showing a main effect of both visual attentional load (F = 123.86, p < 0.01) and of audible noise level (F = 56.45, p < 0.01), and that these effects were additive, with no significant interaction (F = 0.27, p = 0.77). Follow-up t-tests showed that responses were slower under conditions of higher attentional load and noise. Because RTs were not normally distributed, this pattern of results was also verified in median RT calculations.
In an effort to validate that stimulus manipulations indeed had an effect on task difficulty, performance on the visual tasks were also assessed. Mean accuracies were calculated for both the medium and high visual attentional load conditions, with and without audible noise (see Figure 3A). A 2 × 2 ANOVA showed a significant main effect of task (F = 168.46, p < 0.01) and of audible noise level (F = 7.64, p = 0.01). This analysis revealed no interaction between these two factors (F = 0.20, p = 0.66). Follow-up t-tests showed that participants performed as expected, with lower accuracies on the high relative to the low attentional load task, and with lower accuracies in the conditions with noise relative to those without. These effects were additive with simultaneous increase in attentional load and noise.
An analysis of mean RTs provided similar findings (see Figure 3B), with a 2 × 2 ANOVA showing a significant main effect of visual attentional load (F = 462.91, p < 0.01), a marginally significant main effect of audible noise level (F = 4.04, p = 0.053), and a significant interaction between these factors (F = 5.47, p = 0.03). Follow-up t-tests showed that responses were slower with higher attentional loads and noise. This pattern of results was also verified in median RT calculations.
The results from this study show that visual attentional load has a substantial impact on anesthesiologists’ abilities to detect audible changes in oxygen saturation levels. Furthermore, these effects are compounded by the presence of significant noise in the operating room. It must be emphasized that the impact of visual attentional load and audible noise were substantial, with a 17% decrement in accurately-detecting pitch changes between the easiest condition (low visual attentional load in quiet) and the most difficult (high visual attentional load with auditory noise) condition. These deficits were seen not only in the ability to accurately detect a change in oxygen saturation, but also in the speed of reaction when a change was detected. An additional point of emphasis here is that even the most difficult condition in this laboratory setting undoubtedly greatly underestimates the complexity and challenges of a real world operating room.
Given that the majority of anesthesia-related accidents are derivative of compounding small errors10, such as not detecting a change in oxygen saturation, improving such monitoring performance may lead to reduced accident rates. Reducing environmental factors that lead to increased errors should be an important priority for increasing the safety of operating room environments. Specifically, efforts to reduce distracters and lower background noise should be considered during induction and emergence, periods of intense concentration for anesthesiologists 18 and during which they are required to further divide their attention and are thus susceptible to higher rates of error 3–5, 10.
One of the primary critiques of vigilance research is that findings in laboratory settings do not always transfer to the real world 19. While this study did take place in a laboratory setting, two points should be highlighted here. The first is that these participants were responding to simulated pulse oximetry tones as opposed to arbitrary visual cues in previous studies5, 20, an important point given known differences between responses to visual and auditory monitoring 21. Second, all participants were anesthesiology residents who had been trained in the use of such auditory cues. As such, we are confident that these results are of strong relevance for real-world performance with pulse oximetry. In fact, the negative impacts that divided attention and noise have on pulse oximetry measured here are in all probability conservative given that the visual tasks underestimate the complexity of the operating room, and the noise level (mean volume = 67 dB SPL) is significantly lower than the average noise level in the operating room, previously measured at 77 dB SPL 1, 2. With that said, the visual task used here requires constant attention, which is not the case in all stages of clinical anesthesia care, and as such could be viewed as less conservative that a real-world operating room situation. These results are consistent with two previous studies in an operating room setting. In these studies, anesthesiologists were asked to detect a change in artificial visual outputs on a monitor (e.g. the number “5” changing to a “10”). It was found that this signal was missed more often during induction relative to emergence, and emergence relative to maintenance5, 20. These studies thus provide converging evidence that time periods of high attentional load, such as induction and emergence, are associated with increased distraction and less attention afforded to monitors such as pulse oximeters.
These current results illustrate that the increase in error rates with high attentional load and environmental noise are additive, yet it is unknown how these factors relate to additional factors known to influence error rates. Future work should be conducted to explore the relationship between the currently studied factors and those that are clearly important from a performance perspective, including fatigue, sleep deprivation, stress, interpersonal factors, and alarm fatigue 3, 22, 23. Furthermore, future effort should be put forth to explore ways in which anesthesiologists may improve their ability to attend to multiple stimulus inputs across sensory modalities. Specifically, these efforts should utilize research in the field of multisensory processing which has previously investigated the roles of attention and noise on perceptual processing, and which has recently shown the ability of sensory training protocols to enhance sensory performance 24. Given the significant performance decline measured here with attentional demands and noise, both factors that are ubiquitous in the operating room setting, these avenues of research have the capacity to lower error rates and improve patient outcomes.
High visual attentional load impacts the ability of anesthesiologists to detect changes in oxygen saturation levels through pulse oximetry and reduces the speed of response to such changes. These effects are compounded in noisy environments.
Ryan Stevenson: National Institute of Health 1F32 DC011993 (National Institute for Deafness and Communicative Disorders), Bethesda, MD
Joseph Schlesinger: Vanderbilt Institute for Clinical and Translational Research VR2236(Nashville, TN)
Mark Wallace: Vanderbilt Brain Institute(Nashville, TN)
The project described was supported by the National Center for Research Resources, Grant UL1 RR024975-01, and is now at the National Center for Advancing Translational Sciences, Grant 2 UL1 TR000445-06. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Health(Bethesda, MD).