Effectiveness of enhanced pulse oximetry sonifications for conveying oxygen saturation ranges: a laboratory comparison of five auditory displays

Paterson, E; Sanderson, P M; Paterson, N A B; Loeb, R G

doi:10.1093/bja/aex343

Abstract

Background

Anaesthetists monitor auditory information about a patient’s vital signs in an environment that can be noisy and while performing other cognitively demanding tasks. It can be difficult to identify oxygen saturation (⁠ $S p_{O_{2}}$ ⁠) values using existing pulse oximeter auditory displays (sonifications).

Methods

In a laboratory setting, we compared the ability of non-clinician participants to detect transitions into and out of an $S p_{O_{2}}$ target range using five different sonifications while they performed a secondary distractor arithmetic task in the presence of background noise. The control sonification was based on the auditory display of current pulse oximeters and comprised a variable pitch with an alarm. The four experimental conditions included an Alarm Only condition, a Variable pitch only condition, and two conditions using sonifications enhanced with additional sound dimensions. Accuracy to detect $S p_{O_{2}}$ target transitions was the primary outcome.

Results

We found that participants using the two sonifications enhanced with the additional sound dimensions of tremolo and brightness were significantly more accurate (83 and 96%, respectively) at detecting transitions to and from a target $S p_{O_{2}}$ range than participants using a pitch only sonification plus alarms (57%) as implemented in current pulse oximeters.

Conclusions

Enhanced sonifications are more informative than conventional sonification. The implication is that they might allow anaesthetists to judge better when desaturation decreases below, or returns to, a target range.

Editor’s key points

$S p_{O_{2}}$ levels are commonly monitored by sound alterations, also known as sonifications.
The auditory identification of $S p_{O_{2}}$ levels may however be difficult in a noisy and demanding environment.
In this study, $S p_{O_{2}}$ sonifications were enhanced with tremolo or brightness as additional sound dimensions to improve accurate detection of $S p_{O_{2}}$ transitions.
Enhancement of $S p_{O_{2}}$ sonifications resulted in a more accurate recognition of transitions in $S p_{O_{2}}$ levels, and a better detection of desaturation in a noisy environment.

The pulse oximeter auditory display indicates a patient’s heart rate, rhythm, and oxygen saturation (⁠

S p_{O_{2}}

⁠) value when the visual display is not readily available; for example, when an anaesthetist is engaged in other visually demanding tasks or when the visual display is obscured. However, clinicians have difficulty identifying specific

S p_{O_{2}}

values from the auditory tone alone.¹ Furthermore, as task load and noise increase, anaesthetists are less able to distinguish changes in

S p_{O_{2}}

values using only the auditory display.² The situation is exacerbated by the fact that the acoustic properties of current pulse oximeters are not standardized across makes and models.³

Recent research shows that listeners can distinguish designated $S p_{O_{2}}$ ranges and transitions into and out of a target range more accurately with an enhanced auditory display (sonification) than with a variable pitch only sonification.^4–6 The enhanced sonifications in these studies comprised variable pitch with the addition of other sound characteristics in non-target ranges. A limitation of these studies was that participants had no other tasks apart from judging $S p_{O_{2}}$ ranges and detecting transitions into and out of the target range, whereas during surgery anaesthetists perform many tasks in addition to monitoring a patient’s state. Another limitation was that the research was conducted in a quiet room without background noise, whereas noise levels in the operating theatre average from 51 to 75 dB⁷ and can reach values >120 dB.⁸

In this study, we compared participants’ ability to distinguish transitions into and out of a $S p_{O_{2}}$ target range using five different sonifications while performing a distractor task and in the presence of background noise. We predicted that participants using sonifications enhanced with additional sound dimensions would detect $S p_{O_{2}}$ transitions into and out of a predefined target range more accurately than participants using a conventional variable-pitch-only sonification.

Methods

The study was approved by the School of Psychology, The University of Queensland (UQ; ethical clearance number 16-PSYCH-PHD-17_TS). Participants signed an informed consent form agreeing to participate. All data were de-identified.

Power analysis

We conducted a power analysis based on a pilot study that tested 20 participants’ accuracy to detect transitions into and out of a target range using a variable pitch plus alarm sonification and an enhanced variable pitch for multiple patient monitoring sonification. The sample size required to obtain significance for four contrasts at α = 0.05 was ≥14 participants in each condition. We set our number at 20 participants per condition.

Participants

Participants were UQ undergraduate students who were not clinicians or clinical trainees. They received either course credit or a $20 gift voucher. Participants were excluded if (i) they reported hearing abnormalities, or (ii) they had participated in another of our auditory experiments. We allocated each participant to a condition using a block randomization process to ensure equal proportions of musically trained participants in each condition. Musical training was defined as having >1 yr of formal music training.⁹

Design

The experiment was a between-subjects design with one independent variable (sonification) with five conditions: (i) Alarm Only (AO); (ii) Variable pitch (Vp); (iii) Variable pitch plus Alarm (Vp+A); (iv) Enhanced Variable pitch for Single patient monitoring (EVp:S); and (v) Enhanced Variable pitch for Multiple patient monitoring (EVp:M).

In all conditions, there were 30 trials of 60 s duration. In each trial, participants monitored the assigned sonification while $S p_{O_{2}}$ ranged over one or more predefined ranges: Target (100–97%), Low (96–90%), or Critical (89–80%). Heart rate was steady at 72 beats min⁻¹, to correspond to an average adult heart rate. No visual readout of $S p_{O_{2}}$ range or absolute value was present during trials. Participants performed a forced-pace distractor task (arithmetic) throughout each trial.

The primary outcome measure was participants’ accuracy at detecting transitions from the Target to the Low range, or vice versa. Secondary outcome measures were as follows: (i) average latency of detecting transitions between the Target and Low ranges, or vice versa; (ii) accuracy at identifying the $S p_{O_{2}}$ range (Target, Low, or Critical) at the end of each trial; (iii) accuracy at identifying absolute $S p_{O_{2}}$ percentage value (80–100%, plus or minus 1%) at the end of each trial; (iv) accuracy at classifying arithmetic expressions as true or false; (v) average latency of classifying arithmetic expressions as true or false; and (vi) number of arithmetic questions answered. Secondary outcome measures were exploratory in nature.

Apparatus and stimuli

The experiment was conducted on a MacBook Air laptop computer with a 13-inch screen (Apple Computer, Cupertino, CA, USA). Trials were presented using custom software written in Java (Java SE Development Kit, Version jdk1.8.0_40.jdk, Oracle, Redwood Shores, CA, USA). Participants responded using a mouse and keyboard. Sonifications were presented through two active 40 W studio monitor speakers (Behringer MS40, Kirchardt, Germany). The background noise was played through an iPad Mini (Apple Computer).

Sonification conditions

The sound characteristics of the sonifications are in Table 1 (see Supplementary Material to hear the sonifications used). In the Alarm Only (AO) condition, there was no pulse oximetry sound, but an alarm sounded [IEC-Medium-General alarm (IEC-60601-1-8)] ¹⁰ as soon as $S p_{O_{2}}$ entered the Critical range from the Low range and every 15 s thereafter that $S p_{O_{2}}$ remained in the Critical range. If the trial started with $S p_{O_{2}}$ in the Critical range, the alarm first sounded on the second or third pulse tone. This condition corresponds to a typical use of pulse oximetry in the intensive care unit, with the variable pitch tone silenced to reduce noise levels, but with alarms enabled.

Table 1

Sound properties of sonifications. $S p_{O_{2}}$ ⁠, oxygen saturation

Sonification	$S p_{O_{2}}$ range (%)	Sound effects	Sound qualities
Alarm Only (AO)
Target	100–97	–	Silent
Low	96–90	–	Silent
Critical	89–80	IEC-60601-1-8 Medium-General Alarm	Sounds at transition from Low to Critical and every 15 s while $S p_{O_{2}}$ is Critical
Variable pitch (Vp)
Target	100–97	Variable pitch	Pure sine wave
Low	96–90	Variable pitch	Pure sine wave
Critical	89–80	Variable pitch	Pure sine wave
Variable pitch + Alarm (Vp+A)
Target	100–97	Variable pitch	Pure sine wave
Low	96–90	Variable pitch	Pure sine wave
Critical	89–80	Variable pitch plus IEC-60601-1-8 Medium-General Alarm	Pure sine wave + alarm (sounds at transition from Low to Critical and every 15 s while $S p_{O_{2}}$ is Critical)
Enhanced Variable Pitch for Single patient monitoring (EvP:S)
Target	100–97	Variable pitch	Pure sine wave
Low	96–90	Variable pitch + tremolo	Sine wave + four cycles of tremolo per tone (90% wet)
Critical	89–80	Variable pitch + tremolo + brightness	Sine wave + four cycles of tremolo per tone (90% wet) + third, fifth, and seventh harmonics
Enhanced Variable Pitch for Multiple patient monitoring (EvP:M)
Target	100–97	‘All’s well’ sound	Chirp
Low	96–90	Variable pitch + tremolo	Sine wave + four cycles of tremolo per tone (90% wet)
Critical	89–80	Variable pitch + tremolo + brightness	Sine wave + four cycles of tremolo per tone (90% wet) + third, fifth, and seventh harmonics

Sonification	$S p_{O_{2}}$ range (%)	Sound effects	Sound qualities
Alarm Only (AO)
Target	100–97	–	Silent
Low	96–90	–	Silent
Critical	89–80	IEC-60601-1-8 Medium-General Alarm	Sounds at transition from Low to Critical and every 15 s while $S p_{O_{2}}$ is Critical
Variable pitch (Vp)
Target	100–97	Variable pitch	Pure sine wave
Low	96–90	Variable pitch	Pure sine wave
Critical	89–80	Variable pitch	Pure sine wave
Variable pitch + Alarm (Vp+A)
Target	100–97	Variable pitch	Pure sine wave
Low	96–90	Variable pitch	Pure sine wave
Critical	89–80	Variable pitch plus IEC-60601-1-8 Medium-General Alarm	Pure sine wave + alarm (sounds at transition from Low to Critical and every 15 s while $S p_{O_{2}}$ is Critical)
Enhanced Variable Pitch for Single patient monitoring (EvP:S)
Target	100–97	Variable pitch	Pure sine wave
Low	96–90	Variable pitch + tremolo	Sine wave + four cycles of tremolo per tone (90% wet)
Critical	89–80	Variable pitch + tremolo + brightness	Sine wave + four cycles of tremolo per tone (90% wet) + third, fifth, and seventh harmonics
Enhanced Variable Pitch for Multiple patient monitoring (EvP:M)
Target	100–97	‘All’s well’ sound	Chirp
Low	96–90	Variable pitch + tremolo	Sine wave + four cycles of tremolo per tone (90% wet)
Critical	89–80	Variable pitch + tremolo + brightness	Sine wave + four cycles of tremolo per tone (90% wet) + third, fifth, and seventh harmonics

Table 1

Sound properties of sonifications. $S p_{O_{2}}$ ⁠, oxygen saturation

Sonification	$S p_{O_{2}}$ range (%)	Sound effects	Sound qualities
Alarm Only (AO)
Target	100–97	–	Silent
Low	96–90	–	Silent
Critical	89–80	IEC-60601-1-8 Medium-General Alarm	Sounds at transition from Low to Critical and every 15 s while $S p_{O_{2}}$ is Critical
Variable pitch (Vp)
Target	100–97	Variable pitch	Pure sine wave
Low	96–90	Variable pitch	Pure sine wave
Critical	89–80	Variable pitch	Pure sine wave
Variable pitch + Alarm (Vp+A)
Target	100–97	Variable pitch	Pure sine wave
Low	96–90	Variable pitch	Pure sine wave
Critical	89–80	Variable pitch plus IEC-60601-1-8 Medium-General Alarm	Pure sine wave + alarm (sounds at transition from Low to Critical and every 15 s while $S p_{O_{2}}$ is Critical)
Enhanced Variable Pitch for Single patient monitoring (EvP:S)
Target	100–97	Variable pitch	Pure sine wave
Low	96–90	Variable pitch + tremolo	Sine wave + four cycles of tremolo per tone (90% wet)
Critical	89–80	Variable pitch + tremolo + brightness	Sine wave + four cycles of tremolo per tone (90% wet) + third, fifth, and seventh harmonics
Enhanced Variable Pitch for Multiple patient monitoring (EvP:M)
Target	100–97	‘All’s well’ sound	Chirp
Low	96–90	Variable pitch + tremolo	Sine wave + four cycles of tremolo per tone (90% wet)
Critical	89–80	Variable pitch + tremolo + brightness	Sine wave + four cycles of tremolo per tone (90% wet) + third, fifth, and seventh harmonics

Sonification	$S p_{O_{2}}$ range (%)	Sound effects	Sound qualities
Alarm Only (AO)
Target	100–97	–	Silent
Low	96–90	–	Silent
Critical	89–80	IEC-60601-1-8 Medium-General Alarm	Sounds at transition from Low to Critical and every 15 s while $S p_{O_{2}}$ is Critical
Variable pitch (Vp)
Target	100–97	Variable pitch	Pure sine wave
Low	96–90	Variable pitch	Pure sine wave
Critical	89–80	Variable pitch	Pure sine wave
Variable pitch + Alarm (Vp+A)
Target	100–97	Variable pitch	Pure sine wave
Low	96–90	Variable pitch	Pure sine wave
Critical	89–80	Variable pitch plus IEC-60601-1-8 Medium-General Alarm	Pure sine wave + alarm (sounds at transition from Low to Critical and every 15 s while $S p_{O_{2}}$ is Critical)
Enhanced Variable Pitch for Single patient monitoring (EvP:S)
Target	100–97	Variable pitch	Pure sine wave
Low	96–90	Variable pitch + tremolo	Sine wave + four cycles of tremolo per tone (90% wet)
Critical	89–80	Variable pitch + tremolo + brightness	Sine wave + four cycles of tremolo per tone (90% wet) + third, fifth, and seventh harmonics
Enhanced Variable Pitch for Multiple patient monitoring (EvP:M)
Target	100–97	‘All’s well’ sound	Chirp
Low	96–90	Variable pitch + tremolo	Sine wave + four cycles of tremolo per tone (90% wet)
Critical	89–80	Variable pitch + tremolo + brightness	Sine wave + four cycles of tremolo per tone (90% wet) + third, fifth, and seventh harmonics

The Variable pitch (Vp) pulse tones were pure sine wave functions ranging logarithmically from 150 Hz at 80% $S p_{O_{2}}$ to 950 Hz at 100% $S p_{O_{2}}$ ⁠; each 1% change in $S p_{O_{2}}$ corresponded to a 1.84% change in frequency.³ Each tone was 150 ms in duration and included 10 ms fade-in and 10 ms fade-out to eliminate acoustic artifacts. This condition corresponds to use of pulse oximetry sounds when alarm limits are set very wide or alarms have been silenced.

The Variable pitch plus Alarm (Vp+A) condition used the same mapping of pitch to $S p_{O_{2}}$ values as in Vp but included an alarm in the Critical range. As in the AO condition, the alarm sounded as soon as $S p_{O_{2}}$ entered the Critical range and thereafter every 15 s that $S p_{O_{2}}$ remained in the Critical range. This condition corresponds to an intraoperative use of the pulse oximeter with an alarm threshold set at 89%, so it served as the control condition for the present experiment.

In the Enhanced variable pitch for Single patient monitoring (Enhanced Single; EVp:S) condition, the mapping of pitch to $S p_{O_{2}}$ values was as for Vp. In addition, tremolo was added to the pulse tones in the Low and Critical $S p_{O_{2}}$ ranges between 96 and 80%, and brightness was also added in the Critical $S p_{O_{2}}$ range <90%. Tremolo was produced by imposing a sinusoidal amplitude modulation of the pulse tone, four cycles of tremolo with 90% ‘wet’ or depth, resulting in a vibrating effect. Brightness was produced by adding the first three odd harmonics of the fundamental pitch of the pulse tone (i.e. third, fifth, and seventh harmonic) to produce a sharper tone. Previous research showed that adding tremolo in the Low range, and both tremolo and brightness in the Critical range, as described above, improved detection of target transitions and identification of $S p_{O_{2}}$ range.⁶

The Enhanced Variable Pitch for Multiple patient monitoring (Enhanced Multiple; EVp:M) condition was the same as EVp:S condition except that pulse tones were suppressed in the Target range and were replaced by an ‘all’s well’ tone (an acoustic chirp; see below) generated every 5 s while $S p_{O_{2}}$ value remained in the Target range. The 100 ms chirp started with a frequency of 1000 Hz that decreased linearly to 500 Hz at the 50 ms point, and then increased linearly back to 1000 Hz by 100 ms. The volume was initially 0, and increased to 0.3 (on a 0–1 scale) at 50 ms and then decreased linearly to 0 at 100 ms. The chirp was designed using Audacity [Audacity Team (2012), Version 2.02, CMU, Pittsburgh, PA, USA].

Distractor task

The purpose of the distractor task was to impose continuous high load. Participants classified two-digit addition and subtraction arithmetic expressions as ‘true’ or ‘false’ [e.g. 47+22=69 (correct response ‘true’); 37−11=30 (correct response ‘false’)]. Feedback was provided on the screen as ‘correct’ or ‘wrong’. Arithmetic expressions were presented for the full duration of each trial, with equal proportions of true and false instances. Each participant received the expressions in the same order. Performance of the arithmetic task was force paced, with a new expression appearing every 5 s, regardless of whether the participant responded or not.

Background noise

We used Audacity to mix a sound file representing background noise in the operating theatre. It incorporated the following sounds: (i) dialogue; (ii) operating theatre noises, such as suction, clanging of instruments, and a ventilator; and (iii) pop music with vocals. The file was played at a sound level ranging from 45 to 54 dB(A) from the ninth trial in the training block to the end of the experiment.

Questionnaires

Participants completed a pre-experiment questionnaire that asked for their age, gender, and musical training, and whether their hearing was normal. Participants also completed a post-experiment questionnaire probing their reactions to the sonifications (not reported here).

Procedure

We conducted the experiment in a secluded room. The participant sat in front of the laptop computer with the speakers set back ∼15 cm either side of the screen. The iPad Mini playing background noise was placed ∼100 cm behind the participant. We provided each participant with an information sheet and explained the main points of the experiment using a standard protocol. The experimenter emphasized the importance of monitoring blood oxygen values during surgery. Participants were trained to identify the $S p_{O_{2}}$ range, to identify the specific value of $S p_{O_{2}}$ ⁠, and to react when target transitions occurred. The training block comprised 10 trials, with feedback provided throughout the block. After training, the participant completed two experimental blocks of 15 trials each, with trials presented in random order. At the end of each trial, participants worked at their own pace to identify the range and absolute $S p_{O_{2}}$ values. There was a rest period between blocks. Finally, the participant completed the post-experiment questionnaire.

Data analysis

We compared target transition detection accuracy of participants in the Variable pitch plus Alarm (Vp+A) condition with that of participants in the other conditions (AO, Vp, EVp:S, and EVp:M) using P=0.0125 to preserve an α=(0).05 for the primary outcome. Distributions of outcome measures were checked for normality using the Shapiro–Wilk test (P>0.05) and for homogeneity of variance using Levene’s test (P>0.05). If assumptions were met, we conducted parametric tests; specifically, Student’s t-tests. If assumptions were not met, we conducted non-parametric tests; specifically, Mann–Whitney U-tests. SPSS (IBM Corp. IBM SPSS Statistics for MacIntosh, Version 23.0. Armont, NY, USA) was used for all statistical tests.

Results

One hundred participants took part in the experiment. Parametric assumptions of normality and homogeneity of variance were not met, so data were analysed using Mann–Whitney U-tests. Results for the primary outcome, target transition detection accuracy, are shown in Figure 1. Results for patient characteristics and all performance measures are in Table 2.

Table 2

Performance results. Values are the mean [range] or median (95% confidence intervals). AO, Alarms Only; Vp, Variable pitch; Vp+A, Variable pitch plus Alarm; EVp:S, Enhanced Variable pitch for Single patient monitoring; EVp:M, Enhanced Variable pitch for Multiple patient monitoring; $S p_{O_{2}}$ ⁠, oxygen saturation

	AO	Vp	Vp + A (control)	EVp:S	EVp:M
n	20	20	20	20	20
Proportion female (%)	70	45	60	65	60
Age (yr)	23 [18–45]	21[18–25]	22 [18–52]	21 [18–25]	22 [19–28]
Primary outcome
Target transition detection accuracy (%)	42 (40, 43)	67 (57, 70)	57 (53, 63)	83 (77; 93)	96 (90, 100)
P-value	<0.001	0.383		<0.001	<0.001
Secondary outcomes
Target transition detection latency (s)	10.6 (8.1, 17.4)	8.9 (6.6, 12.4)	7.2 (4.7; 9.6)	1.3 (1.0, 2.8)	1.1 (0.9, 1.3)
P-value	0.015	0.231		<0.001	<0.001
Range identification accuracy (%)	60 (57, 67)	60 (50, 67)	75 (70, 87)	87 (83, 93)	90 (87, 93)
P-value	<0.001	<0.001		0.009	<0.001
Absolute $S p_{O_{2}}$ value identification accuracy (%)	28 (23, 33)	32 (27, 40)	48 (43, 57)	65 (50, 73)	65 (57, 70)
P-value	<0.001	<0.001		0.028	0.001
Arithmetic accuracy (%)	80 (70, 85)	75 (68, 81)	73 (67, 79)	77 (71, 80)	84 (72, 87)
P-value	0.165	0.583		0.327	0.072
Arithmetic latency (s)	2.2 (2.0, 2.6)	2.4 (1.9, 2.9)	2.2 (1.9, 2.8)	2.4 (2.1, 2.6)	2.5 (2.1, 2.8)
P-value	0.904	0.883		0.862	0.355
Arithmetic responding (%)	97 (95, 99)	96 (89, 99)	96 (90, 98)	96 (91, 98)	96 (90, 98)
P-value	0.327	0.640		0.947	0.989

	AO	Vp	Vp + A (control)	EVp:S	EVp:M
n	20	20	20	20	20
Proportion female (%)	70	45	60	65	60
Age (yr)	23 [18–45]	21[18–25]	22 [18–52]	21 [18–25]	22 [19–28]
Primary outcome
Target transition detection accuracy (%)	42 (40, 43)	67 (57, 70)	57 (53, 63)	83 (77; 93)	96 (90, 100)
P-value	<0.001	0.383		<0.001	<0.001
Secondary outcomes
Target transition detection latency (s)	10.6 (8.1, 17.4)	8.9 (6.6, 12.4)	7.2 (4.7; 9.6)	1.3 (1.0, 2.8)	1.1 (0.9, 1.3)
P-value	0.015	0.231		<0.001	<0.001
Range identification accuracy (%)	60 (57, 67)	60 (50, 67)	75 (70, 87)	87 (83, 93)	90 (87, 93)
P-value	<0.001	<0.001		0.009	<0.001
Absolute $S p_{O_{2}}$ value identification accuracy (%)	28 (23, 33)	32 (27, 40)	48 (43, 57)	65 (50, 73)	65 (57, 70)
P-value	<0.001	<0.001		0.028	0.001
Arithmetic accuracy (%)	80 (70, 85)	75 (68, 81)	73 (67, 79)	77 (71, 80)	84 (72, 87)
P-value	0.165	0.583		0.327	0.072
Arithmetic latency (s)	2.2 (2.0, 2.6)	2.4 (1.9, 2.9)	2.2 (1.9, 2.8)	2.4 (2.1, 2.6)	2.5 (2.1, 2.8)
P-value	0.904	0.883		0.862	0.355
Arithmetic responding (%)	97 (95, 99)	96 (89, 99)	96 (90, 98)	96 (91, 98)	96 (90, 98)
P-value	0.327	0.640		0.947	0.989

Table 2

Performance results. Values are the mean [range] or median (95% confidence intervals). AO, Alarms Only; Vp, Variable pitch; Vp+A, Variable pitch plus Alarm; EVp:S, Enhanced Variable pitch for Single patient monitoring; EVp:M, Enhanced Variable pitch for Multiple patient monitoring; $S p_{O_{2}}$ ⁠, oxygen saturation

	AO	Vp	Vp + A (control)	EVp:S	EVp:M
n	20	20	20	20	20
Proportion female (%)	70	45	60	65	60
Age (yr)	23 [18–45]	21[18–25]	22 [18–52]	21 [18–25]	22 [19–28]
Primary outcome
Target transition detection accuracy (%)	42 (40, 43)	67 (57, 70)	57 (53, 63)	83 (77; 93)	96 (90, 100)
P-value	<0.001	0.383		<0.001	<0.001
Secondary outcomes
Target transition detection latency (s)	10.6 (8.1, 17.4)	8.9 (6.6, 12.4)	7.2 (4.7; 9.6)	1.3 (1.0, 2.8)	1.1 (0.9, 1.3)
P-value	0.015	0.231		<0.001	<0.001
Range identification accuracy (%)	60 (57, 67)	60 (50, 67)	75 (70, 87)	87 (83, 93)	90 (87, 93)
P-value	<0.001	<0.001		0.009	<0.001
Absolute $S p_{O_{2}}$ value identification accuracy (%)	28 (23, 33)	32 (27, 40)	48 (43, 57)	65 (50, 73)	65 (57, 70)
P-value	<0.001	<0.001		0.028	0.001
Arithmetic accuracy (%)	80 (70, 85)	75 (68, 81)	73 (67, 79)	77 (71, 80)	84 (72, 87)
P-value	0.165	0.583		0.327	0.072
Arithmetic latency (s)	2.2 (2.0, 2.6)	2.4 (1.9, 2.9)	2.2 (1.9, 2.8)	2.4 (2.1, 2.6)	2.5 (2.1, 2.8)
P-value	0.904	0.883		0.862	0.355
Arithmetic responding (%)	97 (95, 99)	96 (89, 99)	96 (90, 98)	96 (91, 98)	96 (90, 98)
P-value	0.327	0.640		0.947	0.989

	AO	Vp	Vp + A (control)	EVp:S	EVp:M
n	20	20	20	20	20
Proportion female (%)	70	45	60	65	60
Age (yr)	23 [18–45]	21[18–25]	22 [18–52]	21 [18–25]	22 [19–28]
Primary outcome
Target transition detection accuracy (%)	42 (40, 43)	67 (57, 70)	57 (53, 63)	83 (77; 93)	96 (90, 100)
P-value	<0.001	0.383		<0.001	<0.001
Secondary outcomes
Target transition detection latency (s)	10.6 (8.1, 17.4)	8.9 (6.6, 12.4)	7.2 (4.7; 9.6)	1.3 (1.0, 2.8)	1.1 (0.9, 1.3)
P-value	0.015	0.231		<0.001	<0.001
Range identification accuracy (%)	60 (57, 67)	60 (50, 67)	75 (70, 87)	87 (83, 93)	90 (87, 93)
P-value	<0.001	<0.001		0.009	<0.001
Absolute $S p_{O_{2}}$ value identification accuracy (%)	28 (23, 33)	32 (27, 40)	48 (43, 57)	65 (50, 73)	65 (57, 70)
P-value	<0.001	<0.001		0.028	0.001
Arithmetic accuracy (%)	80 (70, 85)	75 (68, 81)	73 (67, 79)	77 (71, 80)	84 (72, 87)
P-value	0.165	0.583		0.327	0.072
Arithmetic latency (s)	2.2 (2.0, 2.6)	2.4 (1.9, 2.9)	2.2 (1.9, 2.8)	2.4 (2.1, 2.6)	2.5 (2.1, 2.8)
P-value	0.904	0.883		0.862	0.355
Arithmetic responding (%)	97 (95, 99)	96 (89, 99)	96 (90, 98)	96 (91, 98)	96 (90, 98)
P-value	0.327	0.640		0.947	0.989

Fig 1

Target transition detection accuracy for all sonification conditions (medians and non-parametric 95% confidence intervals). Vp+A is the control condition (Variable pitch plus Alarm). *Significantly different from Vp+A; ns denotes not significantly different from Vp+A.

Open in new tab Download slide

Primary outcome

Participants in the enhanced conditions (EVp:S and EVp:M) detected target transitions significantly more accurately than participants in the Variable pitch plus Alarm (Vp+A) condition (P<0.001 in both instances). Participants in the Alarm Only (AO) condition identified target transitions significantly less accurately than participants in the Vp+A condition (P<0.001), but there was no difference between the Variable pitch (Vp) and Vp+A conditions (P=0.383).

Secondary outcomes

Participants in the EVp:S and EVp:M conditions were significantly faster at detecting target transitions than participants in the Vp+A condition (P<0.001 in both instances). Participants in the AO condition were significantly slower at detecting target transitions than participants in the Vp+A condition (P=0.015), but there was no difference between the Vp and Vp+A conditions (P=0.231).

Participants in the EVp:S and EVp:M conditions identified $S p_{O_{2}}$ range significantly more accurately than participants in the Vp+A condition (P=0.009 and P<0.001, respectively). Participants in the AO and Vp conditions were significantly less accurate at identifying range than those in the Vp+A condition (P<0.001 in both instances).

Participants in the EVp:S and EVp:M conditions identified the absolute $S p_{O_{2}}$ percentage value significantly more accurately than participants in the Vp+A condition (P=0.028 and P<0.001, respectively). Participants in the AO and Vp conditions identified the absolute $S p_{O_{2}}$ value significantly less accurately than those in the Vp+A condition (P<0.001 in both instances).

As a manipulation check, we compared participants’ arithmetic accuracy, latency, and number of responses, and found no significant differences between the Vp+A condition and any of the other four conditions.

Discussion

Despite wide acceptance of pulse oximetry as essential for anaesthesia care and despite its mandate as standard equipment in operating theatres in many countries,¹¹ the variable pitch sonification does not always provide important patient information effectively.³^,¹² Although there is evidence that current pulse oximeter sonifications are effective for informing clinicians when saturation decreases rapidly, remains relatively stable, or increases rapidly, there is mounting evidence that they cannot judge absolute $S p_{O_{2}}$ values using the current auditory displays.¹^,²^,¹² Adding further sound properties to attract auditory attention to clinically important thresholds was suggested by Watson and Sanderson,¹³ and recent laboratory investigations demonstrate the potential of this method for enhancing pulse oximeter sonifications.^4–6 However, participants in the recent studies focused solely on the task of $S p_{O_{2}}$ monitoring in a quiet environment, and there were no alarms used in the control condition.

In the present study, participants listening to two enhanced sonifications (EVp:S and EVp:M) could detect transitions into and out of an $S p_{O_{2}}$ target range more accurately than participants listening to a control Variable pitch sonification plus Alarm (83 and 96 vs 57%, respectively). Accuracy rates with the Enhanced Multiple sonification approached 100%, even though participants performed a demanding distractor task in the presence of representative operating theatre background noise.^14–16 Secondary outcomes of $S p_{O_{2}}$ range identification and absolute $S p_{O_{2}}$ identification also showed striking patterns of improved performance when using the enhanced sonifications. The additional auditory dimensions in the enhanced sonifications appear to give listeners more information about $S p_{O_{2}}$ range. The presence or absence of tremolo effectively signalled the transition across the boundary between Target and Low ranges. In the Enhanced Multiple condition, presence of the ‘all’s well’ chirp in the Target range had an even more pronounced effect.

Participants in the Alarm Only condition heard only silence when target transitions occurred, because the alarm sounded only within the Critical range. It is therefore not surprising that their performance at detecting target transitions, and identifying range or absolute $S p_{O_{2}}$ values, was worse than for participants in the Variable pitch plus Alarm condition.

There were no differences in how accurately or how quickly participants using the Variable pitch condition detected target transitions when compared with participants using the Variable pitch plus Alarm condition, probably because participants in both conditions heard the same sonification (variable pitch only) for the Target and Low range. Given that the median of the Variable pitch condition was non-significantly higher than for the Variable pitch plus Alarm condition, we conducted a post hoc analysis to check whether participants in the Enhanced Single condition were detecting target transitions more accurately than participants in the Variable pitch condition, and found that this was so (P<0.001). Participants in the Variable pitch condition were significantly less accurate at identifying $S p_{O_{2}}$ range or absolute $S p_{O_{2}}$ value than participants in the Variable pitch plus Alarm condition, probably because the alarm in the latter condition helped participants to discriminate the Low and Critical ranges.

It is an industry requirement to include auditory threshold alarms on pulse oximeters to alert clinicians to critical $S p_{O_{2}}$ values.¹⁷ However, ∼70% of anaesthetists report silencing alarms because they find the high false alarm rate annoying and distracting.¹⁸ Existing sonifications are not designed to capture attention when saturation crosses a clinically relevant threshold; thus, the need for alarms. An effective sonification might enable clinicians to monitor $S p_{O_{2}}$ values pre-attentively while they perform other cognitively engaging tasks, thereby making it possible to take remedial action before critical values are reached and alarms sound. An effective sonification might also let clinicians set wider alarm limits. In both instances, the number of alarms sounding could decrease. Additionally, an enhanced sonification might allow for more accurate monitoring of whether the $S p_{O_{2}}$ has returned to target values after a desaturation event and remedial action.

Limitations

Our rationale for conducting this study in a laboratory setting was to investigate sonifications from the perspective of auditory perception before testing in a more representative operating theatre setting. Therefore, the study has real or potential limitations that make it premature to assume that the enhanced sonifications are ready for the clinical context.

First, of the two enhanced sonifications tested, only the Enhanced Single sonification is a plausible candidate for normal monitoring of one patient at a time. Although performance with Enhanced Multiple sonification was best, it is a prototype for situations where more than one patient’s $S p_{O_{2}}$ might require monitoring. It was developed as part of a programme of research to design sonifications for monitoring multiple patients, where simultaneous sonifications from different patients are impractical. In a multiple patient situation, when the $S p_{O_{2}}$ of each patient is in their target range a single ‘all’s well’ sound would be heard at regular intervals. If the $S p_{O_{2}}$ value were to move out of the target range for any patient, the ‘all’s well’ sound would be replaced by enhanced variable pitch pulse tones for that patient. The present experiment provided a first opportunity to test this concept; however, further design is needed. We do not recommend use of the ‘all’s well’ sound in the target range for normal monitoring of one patient. For the anaesthetized patient, the variable pitch sonification is an important indicator of saturation changes even within the target range.

Second, participants in this experiment were non-clinicians. Anaesthetists’ greater experience of allocating attention across intraoperative tasks and their familiarity with pulse oximeter auditory displays might mean that their ability to identify $S p_{O_{2}}$ parameters is better than non-clinicians using a conventional sonification.¹⁵ However, it would be surprising if anaesthetists were so much better with the conventional sonification that the Enhanced Single sonification was no longer superior.

Third, we did not test participants’ pitch detection capability, but instead asked for their level of formal music training. In previous work, we have found that the presence or absence of formal music training can be a sensitive predictor of participants’ performance with auditory alarms⁹ and sonifications,⁴ but such differences become unimportant if all participants perform above some criterion level with a display. In the present experiment, we used block randomisation to ensure equal proportions of participants with and without formal music training in each condition. The performance of participants with and without formal music training was statistically indistinguishable for all measures.

Fourth, the experiment was carried out in a laboratory setting in the presence of a single experimenter, with a distractor task unrelated to the monitoring task, and with only a recording of background operating theatre sounds playing. The anaesthetist has numerous tasks to perform concurrently in a busy environment filled with social and auditory distractions.¹⁹^,²⁰ In future research, we plan to compare anaesthetists’ ability to detect saturation transitions using enhanced sonifications compared with conventional sonifications during longer scenarios in a high-fidelity simulator.

Conclusions

This study yields further evidence that pulse oximeter sonifications enhanced with additional sound dimensions provide more informative signals for monitoring $S p_{O_{2}}$ than does a variable pitch sonification plus alarms, when participants are engaged in a cognitively demanding task in the presence of background sounds. Enhanced sonifications might offer earlier detection of patient deterioration during anaesthesia, more timely interventions, and more precise assessment of the efficacy of corrective strategies. This study also provides early data on an auditory display that could be used for monitoring $S p_{O_{2}}$ in multipatient contexts. More informative pulse oximetry sonifications have implications not only for anaesthetic practice, but also for monitoring patients in other hospital settings, including intensive care units, emergency medicine departments, and during patient transport.

Authors’ contributions

Conceived the study: P.M.S., R.G.L., E.P.

Reviewed the design of sonification stimuli, helped to compile background noise files, and provided information on operating theatre context: R.G.L., N.A.B.P.

Participant recruitment and data collection: E.P.

Data analysis: P.M.S., E.P.

First draft of the manuscript and revision: E.P.

Critical review and revision: P.M.S., N.A.B.P., R.G.L.

Supplementary material

Supplementary material is available at British Journal of Anaesthesia online.

Acknowledgements

The authors thank Dr Birgit Brecknell for software programming; Anna Hickling for help with creating auditory stimuli; and Caitlin Browning, Anna Hickling, and Jeffrey Kim for their help with recording the background noise file.

Declaration of interest

P.M.S. is co-inventor of a respiratory sonification (Sanderson and Watson, US Patent 7070570).

R.G.L. has received $1000 per year to be on the Masimo, Inc. Scientific Advisory Board.

Funding

Australian Research Council (Discovery Project Grant DP140101822 to P.M.S., R.G.L., and David Liu; and an Australian Postgraduate Award to E.P.).

References

1

Morris

RW

,

Mohacsi

PJ.

How well can anaesthetists discriminate pulse oximeter tones?

Anaesth Intensive Care

2005

;

33

:

497

–

500

Google Scholar

PubMed

OpenURL Placeholder Text

WorldCat

2

Stevenson

RA

,

Schlesinger

JJ

,

Wallace

MT.

Effects of divided attention and operating room noise on perception of pulse oximeter pitch changes: a laboratory study

.

Anesthesiology

2013

;

118

:

376

–

81

3

Loeb

RG

,

Brecknell

B

,

Sanderson

P.

The sounds of desaturation: a survey of commercial pulse oximeter sonifications

.

Anesth Analg

2016

;

122

:

1395

–

403

4

Hinckfuss

K

,

Sanderson

P

,

Loeb

RG

,

Liley

H

,

Liu

D.

Novel pulse oximetry sonifications for neonatal oxygen saturation monitoring: a laboratory study

.

Hum Factors

2016

;

58

:

344

–

59

5

Deschamps

M-L

,

Sanderson

P

,

Hinckfuss

K

, et al.

Improving the detectability of oxygen saturation level targets for preterm neonates: a laboratory test of tremolo and beacon sonifications

.

Appl Ergon

2016

;

56

:

160

–

9

6

Paterson

E

,

Sanderson

P

,

Paterson

NAB

,

Liu

D

,

Loeb

RG.

The effectiveness of pulse oximetry sonification enhanced with tremolo and brightness for distinguishing clinically important oxygen saturation ranges: a laboratory study

.

Anaesthesia

2016

;

71

:

565

–

72

7

Hasfeldt

D

,

Laerkner

E

,

Birkelund

R.

Noise in the operating room—what do we know? A review of the literature

.

J Perianesth Nurs

2010

;

25

:

380

–

6

8

Kracht

JM

,

Busch-Vishniac

IJ

,

West

JE.

Noise in the operating rooms of Johns Hopkins Hospital

.

J Acoust Soc Am

2007

;

121

:

2673

–

80

9

Wee

AN

,

Sanderson

PM.

Are melodic medical equipment alarms easily learned?

Anesth Anal

2008

;

106

:

501

–

8

Google Scholar

Crossref

WorldCat

10

International Electrotechnical Commission

. Medical electrical equipment, part 1-8: General requirements, tests and guidance for alarm systems in medical electrical equipment and medical electrical systems.

2006

. Geneva. IEC 60601-1-8:2006. Available from https://www.iso.org/standard/41986.html (accessed 16 February 2017)

11

Shah

A

,

Shelley

KH.

Is pulse oximetry an essential tool or just another distraction? The role of the pulse oximeter in modern anesthesia care

.

J Clin Monit Comput

2013

;

27

:

235

–

42

12

Schulte

GT

,

Block

FE.

Can people hear the pitch change on a variable-pitch pulse oximeter?

J Clin Monit

1992

;

8

:

198

–

200

13

Watson

MO

,

Sanderson

P.

Designing for attention with sound: challenges and extensions to ecological interface design

.

Hum Factors

2007

;

49

:

331

–

46

14

Gaba

DM

,

Lee

T.

Measuring the workload of the anesthesiologist

.

Anesth Analg

1990

;

71

:

354

–

61

15

Watson

MO

,

Sanderson

P.

Sonification supports eyes-free respiratory monitoring and task time-sharing

.

Hum Factors

2004

;

46

:

497

–

517

16

Sanderson

P

,

Watson

MO

,

Russell

WJ

, et al.

Advanced auditory displays and head-mounted displays: advantages and disadvantages for monitoring by the distracted anesthesiologist

.

Anesth Analg

2008

;

106

:

1787

–

97

17

Checketts

MR

,

Alladi

R

,

Ferguson

K

, et al.

Recommendations for standards of monitoring during anaesthesia and recovery 2015: Association of Anaesthetists of Great Britain and Ireland

.

Anaesthesia

2016

;

71

:

85

–

93

18

Block

FE

Jr,

Nuutinen

L

,

Ballast

B.

Optimization of alarms: a study on alarm limits, alarm sounds, and false alarms, intended to reduce annoyance

.

J Clin Monit Comput

1999

;

15

:

75

–

83

19

Campbell

G

,

Arfanis

K

,

Smith

A.

Distraction and interruption in anaesthetic practice

.

Br J Anaesth

2012

;

109

:

707

–

15

20

Phipps

D

,

Meakin

GH

,

Beatty

PC

,

Nsoedo

C

,

Parker

D.

Human factors in anaesthetic practice: insights from a task analysis

.

Br J Anaesth

2008

;

100

:

333

–

43

Editor:

Download all slides

Month:	Total Views:
October 2017	49
November 2017	130
December 2017	63
January 2018	11
February 2018	7
March 2018	11
April 2018	8
May 2018	3
June 2018	16
July 2018	1
August 2018	15
September 2018	5
October 2018	3
November 2018	1
December 2018	227
January 2019	235
February 2019	238
March 2019	294
April 2019	284
May 2019	264
June 2019	222
July 2019	204
August 2019	216
September 2019	99
October 2019	41
November 2019	24
December 2019	23
January 2020	23
February 2020	8
March 2020	11
April 2020	11
May 2020	11
June 2020	25
July 2020	21
August 2020	13
September 2020	19
October 2020	15
November 2020	5
December 2020	12
January 2021	8
February 2021	13
March 2021	21
April 2021	8
May 2021	9
June 2021	10
July 2021	13
August 2021	16
September 2021	17
October 2021	7
November 2021	5
December 2021	6
January 2022	1
February 2022	5
March 2022	8
April 2022	13
May 2022	4
June 2022	7
July 2022	16
August 2022	17
September 2022	10
October 2022	12
November 2022	6
December 2022	5
January 2023	1
February 2023	2
March 2023	2
April 2023	8
May 2023	4
July 2023	2
August 2023	2
September 2023	3
October 2023	5
November 2023	5
December 2023	7
January 2024	5
February 2024	5
March 2024	2
April 2024	11
May 2024	4
June 2024	2
July 2024	8
August 2024	2
September 2024	5
October 2024	7
November 2024	8
December 2024	3
February 2025	3
March 2025	11
April 2025	3

Article Contents

Effectiveness of enhanced pulse oximetry sonifications for conveying oxygen saturation ranges: a laboratory comparison of five auditory displays

Abstract

Methods

Power analysis

Participants

Design

Apparatus and stimuli

Sonification conditions

Distractor task

Background noise

Questionnaires

Procedure

Data analysis

Results

Primary outcome

Secondary outcomes

Discussion

Limitations

Conclusions

Authors’ contributions

Supplementary material

Acknowledgements

Declaration of interest

Funding

References

Supplementary data

Citations

Views

Altmetric

See also

Companion Article

Citing articles via

Latest

Most Read

Most Cited

Article Contents

Effectiveness of enhanced pulse oximetry sonifications for conveying oxygen saturation ranges: a laboratory comparison of five auditory displays

Abstract

Methods

Power analysis

Participants

Design

Apparatus and stimuli

Sonification conditions

Distractor task

Background noise

Questionnaires

Procedure

Data analysis

Results

Primary outcome

Secondary outcomes

Discussion

Limitations

Conclusions

Authors’ contributions

Supplementary material

Acknowledgements

Declaration of interest

Funding

References

Supplementary data

Citations

Views

Altmetric

See also

Companion Article

Citing articles via

Latest

Most Read

Most Cited

This Feature Is Available To Subscribers Only