Applied Ergonomics xxx (2014) 1e7
Contents lists available at ScienceDirect
Applied Ergonomics journal homepage: www.elsevier.com/locate/apergo
Exploring positive hospital ward soundscape interventions J. Mackrill*, P. Jennings, R. Cain WMG, University of Warwick, Coventry CV4 7AL, UK
a r t i c l e i n f o
a b s t r a c t
Article history: Received 30 July 2013 Accepted 6 April 2014 Available online xxx
Sound is often considered as a negative aspect of an environment that needs mitigating, particularly in hospitals. It is worthwhile however, to consider how subjective responses to hospital sounds can be made more positive. The authors identified natural sound, steady state sound and written sound source information as having the potential to do this. Listening evaluations were conducted with 24 participants who rated their emotional (Relaxation) and cognitive (Interest and Understanding) response to a variety of hospital ward soundscape clips across these three interventions. A repeated measures ANOVA revealed that the ‘Relaxation’ response was significantly affected (n2 ¼ 0.05, p ¼ 0.001) by the interventions with natural sound producing a 10.1% more positive response. Most interestingly, written sound source information produced a 4.7% positive change in response. The authors conclude that exploring different ways to improve the sounds of a hospital offers subjective benefits that move beyond sound level reduction. This is an area for future work to focus upon in an effort to achieve more positively experienced hospital soundscapes and environments. Ó 2014 Elsevier Ltd and The Ergonomics Society. All rights reserved.
Keywords: Soundscape Healthcare Environment
1. Introduction It is commonly accepted that the hospital environment affects patients’ experiences when they are admitted and treated in these spaces (see Dijkstra et al., 2006). Research on healthcare design and planning has highlighted strong relationships between environmental characteristics and human health (Monti et al., 2012). Sound is one part of the hospital environment. Ergonomics evaluations commonly consider the sound level of spaces, such as offices (Passero and Zannin, 2012) and consider the effect of them in reference to guidelines. Many studies have used this approach for hospitals and documented sound level within acute care hospital spaces (Busch-Vishniac et al., 2005; Okcu et al., 2011; Salandina et al., 2011). Sound levels range from 60 to 70 dB(A) with peaks of around 90 dB(A), equivalent to a busy motorway (Joseph and Ulrich, 2007). These exceed recommendations produced by bodies such as the World Health Organisation (Berglund et al., 2000) or the UK Government (Department of Health, 2008) by around 20e30 dB(A). Consideration is needed however, of how sound contributes to the feeling of the hospital space as ‘quietness’ is a quality indicator which is associated with a lack of annoying noises rather than the * Corresponding author. Tel.: þ44 24761 50760. E-mail address: [email protected] (J. Mackrill).
absence of sound itself (Fornara et al., 2006). Although there can be negative health consequences of excessive sound within these spaces (Ulrich, 1992), there is potential for positive benefits in understanding and maintaining the soundscape of the ward; for instance hearing the various occupational sounds, such as tea trolleys, as positive sounds (Mackrill et al., 2013a). A soundscape is defined as the auditory version of a landscape (Schafer, 1976) and considers the complete sound environment rather than focusing on a single source such as the beep of a monitor. As control of hospital sounds should consider positive aspects and move away from sound level measurement (Dawson, 2005) understanding the soundscape offers a new approach, particularly as the sound quality is not simply determined by level but other factors such as physical space (Hignett and Lu, 2010). Investigating the subjective response to the existing soundscape may create a more feasible way of improving the sound of the spaces, particularly as there is a consistent trend of rising levels since the 1960s (Busch-Vishniac et al., 2005). Therefore, a need to evaluate the effectiveness of interventions aiming to improve the subjective response to a hospital ward soundscape has arisen. Although studies have investigated subjective responses to hospital sounds to a certain degree (Topf, 1985; Akansel and Kaymakci, 2008; Xie and Kang, 2010) none have considered how to create a positive soundscape using interventions which may be of practical benefit.
http://dx.doi.org/10.1016/j.apergo.2014.04.005 0003-6870/Ó 2014 Elsevier Ltd and The Ergonomics Society. All rights reserved.
Please cite this article in press as: Mackrill, J., et al., Exploring positive hospital ward soundscape interventions, Applied Ergonomics (2014), http://dx.doi.org/10.1016/j.apergo.2014.04.005
2
J. Mackrill et al. / Applied Ergonomics xxx (2014) 1e7
1.1. Aim
Table 1 Sound clips and broad content classification for the 12 soundscape clip stimuli.
The aim of this study was to use laboratory listening evaluations to test interventions which might create a more positive emotionalcognitive response to hearing the soundscape of a cardiothoracic (CT) ward. This evaluated the effect of soundscape interventions to understand which may be considered appropriate for investigation within the ward setting. This was the third stage in a four part research project aiming at understanding and improving the perception of a hospital ward soundscape (Fig. 1). Part 1 aimed to understand the perception of the soundscape and inform the direction of subsequent steps. Part 2 created a framework to measure the response to the soundscape with this being used to understand the effect of soundscape interventions in Part 3, presented here. From this theoretical generalisation Part 4 was carried out in-situ to evaluate the chosen intervention. 2. Soundscape intervention rationale The following soundscape interventions were used based on the premise that sound level reduction would not necessary create a more positive soundscape. As a result it was hypothesised that the soundscape would elicit more positive responses when each intervention was applied to the existing sound clips of the CT soundscape. 2.1. Natural sound Natural sound (NS) was selected due to the agreement from literature (Guastavino, 2006; Pheasant et al., 2010; Yang and Kang, 2005a,b) which advocates that natural sounds of birdsong and water are positive soundscape features. Additionally, Mackrill et al. (2013a) found natural sounds to be a positive aspect of the CT ward soundscape when heard through windows overlooking a green space. 2.2. Steady state sound Steady state sound (SSS) was chosen as an intervention as masking sounds are a way in which negative sound can be controlled (Bowman, 1974). Defined as the presence of one sound that renders another sound undetectable (Plack, 2005), this has been tested to improve hospital soundscape perception (see Stanchina et al., 2005). When white noise was added to the sounds of an intensive care unit, despite an increase in sound level, subjectively sleep was consolidated and arousal was less frequent (Stanchina et al., 2005). Despite a small sample and a nonsignificant effect, this shows a change in perception.
Clip number
Playback dB(A)95%
Broad classification of dominant sounds
1 2
72.31 65.78
3
63.24
4 5 6
64.67 68.20 68.03
7
63.78
8
67.56
9
62.34
10
65.43
11 12
69.64 68.75
Conversation and sterilising machine sounds Conversation and footsteps. Doors opening and closing with objects banging. Conversation and footsteps in background. Trolley passing with objects banging, doors opening and closing. Quiet corridor. Objects banging in background. Deep rumble of trolley passing. Conversation and footsteps in background. Patient screaming intermittently. Conversation and footsteps in background. Trolley passing. Talking by nurse talking patient observations monitors beeping and sounds of a car from outside. Quiet patient bay with private conversation from nurse to patient in background. Quiet patient bay with monitor beeps, rustling, and talking in background. Floor cleaning buffer, long monitor beeps. Loud nurse conversation with laughter, monitor beeps.
2.3. Sound source information Written information detailing the various sound sources of the soundscape was chosen as hospital wards produce a sensory overload of unfamiliar stimuli which bombard patients with experiences they are unaccustomed to (Akansel and Kaymakci, 2008). Understanding hospital sounds however, enables patients to habituate and accept sound (Mackrill et al., 2013a). As complex environmental sounds are processed as meaningful events, providing information about interactions with environment (Guastavino et al., 2005), it may be necessary for patients to understand the sounds of the ward in order to feel comfortable within it. Furthermore, Topf (2000) suggests that personal control is the capacity to regulate stress caused by a negative event. This may be a cognitive approach by having information about the stressor (Topf, 2000). As such, it was proposed that providing information detailing what can be heard in the soundscape may improve the feeling that the soundscape elicits. This is called sound source information (SSI). 3. Method 3.1. Sound laboratory and stimuli As there is a challenge for clinicians, designers and researchers to work together using robust high quality research methods (Hignett and Lu, 2010) a laboratory listening evaluation method
Fig. 1. Timeline of the research project.
Please cite this article in press as: Mackrill, J., et al., Exploring positive hospital ward soundscape interventions, Applied Ergonomics (2014), http://dx.doi.org/10.1016/j.apergo.2014.04.005
J. Mackrill et al. / Applied Ergonomics xxx (2014) 1e7
was chosen to scope the potential of these interventions. This provided the means to facilitate the investigation of the inventions in an easily repeatable and controlled setting. Evaluations were held in a Sound Room laboratory. This consisted of a 16 speaker system (using KEF iQ70 and HTS3001SE speakers) operated via a PC computer using Nuendo 4 sound editing software. The system played back hospital soundscape recordings through all 16 speakers. Twelve twenty second soundscape excerpts were presented to participants. These were selected from a database of 32 recordings taken within the CT ward using a Bruel and Kjaer binaural SonoScout recording device. The most salient clips were selected for presentation informed by the analysis of all 32 recordings using the coding procedure developed by Poxon et al. (2009) (Table 1). For interventions of NS and SSS, the associated sounds were incorporated into all clips. NS was obtained from the website www. freesoundproject.org [accessed 3rd March 2011] and contained the sound of a blackbird singing and a babbling stream. SSS was obtained from using a short one second recording of a steriliser machine sound present within the CT ward providing a steady nondescript sound. The sound level of these interventions was set so they formed a congruent background feature rather than a dominant foreground sound. These sounds were only played through the ceiling speakers to help achieve this with the researcher judging the sound level at which they were set. SSI was presented on a response sheet (Table 2). This information was designed from the sound sources within the 12 soundscape clips. The information was the same for each soundscape clip rather than specifying details of the individual sources within each of the 12 clips. This was to assess if SSI would work in a generic manner. Further, as the hospital soundscape is constantly varying an in-situ SSI intervention would have to be generic in nature to accommodate the majority of features rather than focusing on specific nuances. 3.2. Experimental procedure Each soundscape clip lasted around 20 s presented sequentially with a 10 s inter stimulus period to discriminate between clips. Participants completed each of the four conditions on separate occasions. Interventions were applied to the 12 soundscape clips during each condition. Evaluations were conducted a minimum of two day apart to avoid demand effects, however this varied according to participant availability. To avoid order effects, participants were split into four groups with the condition sequence order randomised using a Latin square counterbalancing method. In addition, each group started the evaluation on a different soundscape clip, either 1, 2, 3 or 4, to further control for bias. Responses were recorded on a paper questionnaire using four seven point semantic differential scales. These rating scales were derived from the emotional-cognitive dimensions which the authors established in study 2 (see Mackrill et al., 2013b). The perceptual dimensions were labelled ‘Relaxation’ and ‘Interest and Understanding’. ‘Relaxation’ was measured using the bipolar semantic scales ‘relaxed-stressed’ and ‘comfortable-uncomfortable’. ‘Interest and Understanding’ used the scales ‘curious-uninterested’ and ‘intrigued-bored’. At the end of the questionnaire, basic demographic questions and open questions were asked to provide descriptive data beyond numerical quantification and included: Were there any sounds that particularly affected you? How did you feel overall when listening to the recordings? How does having information on the soundscape make you feel?
3
Table 2 SSI as an intervention used to detail the soundscape clip. What you may hear
Associated activities
Blood pressure monitors beeping for observations. Nurses talking to patients about how they feel. Trolleys moving equipment around the ward. Cleaning machines to keep the ward tidy. General bustling of the ward, patients going for treatments, phones ringing etc. Sound of trolleys bringing in food. Jingling of cups. Patients talking and moving around. Other staff talking.
Patient Observations. Cleaning. Bed Changing. Chatting. Washing.
Lunch.
3.3. Analysis The reliability of the scales measuring each dimension was tested across conditions to ensure that the results could be justifiably compared. This was carried out using a Cronbach’s a test. The main effect was analysed using a repeated measures analysis of variance (ANOVA) and was performed using statistical software SPSS 19 once the normality of the data had been established. 3.4. Sample Sample size was calculated using GPower3 software (Faul et al., 2007). Setting a power level of 0.80 with an effect size (eta2) of 0.25 (small) for a repeated measure ANOVA with four conditions and one group, sample size was calculated to be n ¼ 24 (a level set 0.05 resulting in a test power of 0.81, CI 95%). The 24 participants had a mean age of 32 years (S.D. 10.13 yrs), with 13 male and 11 female. The sample included a range of non-expert demographics from within and outside the University community. Recruitment took place using a convenience sampling strategy. Although a sample demographic was obtained that was removed from those specific to the CT ward, the soundscape of a CT ward was chosen due to the variety of sound representative of general hospital spaces. Therefore the sample demographic may experience such soundscapes which supported this strategy. Further this study was used to inform theoretical rather than empirical generalisation: generation of theoretical concepts deemed to be of wider application (Ritchie and Lewis, 2003). The exploration of empirical generalisation was carried out in stage 4 of the research project. 4. Results1 4.1. Scale reliability The ‘Relaxation’ dimension recorded high reliability, Cronbach’s
a ¼ 0.921 across all conditions. Likewise ‘Interest and Understanding’ recorded Cronbach’s a ¼ 0.895, suggesting that the scales were consistent in measuring the emotional-cognitive response and thus allowing a valid comparison between conditions. 4.2. Repeated measures ANOVA main effect The results showed a significant difference across all conditions on the ‘Relaxation’ dimension; (F (3,751) ¼ 13.991, p ¼ 0.001). A small overall effect size of 0.05 (partial h2) showed that 5% of the variation in emotional-cognitive response can be accounted for by
1 All analysis was performed on the combined mean scores for each dimension i.e. ‘Relaxation’ was measured using the mean scores of scales stressed-relaxed and comfortable-uncomfortable; ‘Interest and Understanding’ using the mean score of scales curious-uninterested and intrigued-bored.
Please cite this article in press as: Mackrill, J., et al., Exploring positive hospital ward soundscape interventions, Applied Ergonomics (2014), http://dx.doi.org/10.1016/j.apergo.2014.04.005
4
J. Mackrill et al. / Applied Ergonomics xxx (2014) 1e7
Private conversation between patients and nurses I don’t feel I should be listening to these. (P12) The crying lady [sound clip 6, a patient with dementia crying out] was upsetting. (P3) [I] loath beeping noise. Grates a bit. (P13) Mixture of boredom and curiosity to generally unpleasant soundscape. (P24) Fluid bags didn’t sound nice when rustling. (P5)
Control
7
Natural Sound Steady State Sound 6
Sound Source Information
5 Relaxation 1
2
4
3
4
5
6
7
The addition of NS was described as: 3
Relaxing. (P3) I find hospital make me feel claustrophobic. The birdsong and water make it feel less so like being next to a window. (P19) Bought a soothing dimension. (P24)
2
1 Interest & Understanding
However, despite positive comments some were negative:
Fig. 2. The effect of each intervention on each clip mapped within the twodimensional space using the mean scores for each condition. Note, increase in ‘Relaxation’ in interventions with the highest NS and SSI.
the differing conditions. A non-significant difference was seen between conditions on the ‘Interest and Understanding’ dimension (F(3,764) ¼ 1.447, p ¼ 0.229) (Fig. 2). Pairwise comparisons, with a Bonferroni correction to account for the increase in pairwise comparisons, showed a significant difference between the control and all interventions (Table 3). NS caused the largest change in response; 10.1%; (mean difference NS ¼ 0.445, p ¼ 0.001), CI (95%) 0.249e0.637). SSS had a smaller difference causing a 3.3% change; (mean difference SSS ¼ 0.208, p ¼ 0.008. CI (95%) 0.038e0.378). Finally, SSI was successful in producing a moderate difference in scores almost half that of natural sound with 4.7%; (mean difference SSI ¼ 0.247, p ¼ 0.001), CI (95%) 0.089e0.406). Pairwise comparisons for ‘Interest and Understanding’ explored the smaller difference between the control condition and interventions (Table 3). The mean difference of NS ¼ 0.00, p ¼ 1.00 corresponding to a 0.6% change. However, SSS caused a larger nonsignificant change of 4.7%, mean difference SSS ¼ 0.123, p ¼ 0.338. Finally, SSI produce a small effect mean difference ¼ 0.066, p ¼ 1.00, resulting in a 2% change. 4.3. Subjective comments Although the manner of each intervention was not divulged to participants, subjective response to the conditions showed trends in response to positive and negative sounds. Negative sounds included patient and staff conversation, particularly when private. Sounds of monitor alarms were also described as annoying:
I like the birdsong but it could get too much if ‘piped’ in. (P3) With birds and conversation I felt life was passing me by. (P18) Sound of running water didn’t fit. (P6) To be left in quiet all day would drive me just as mad as listening to alarms. (P8)
The result of SSS was considered more negative with no positive free responses obtained: White noise and beeps don’t make me relaxed. (P8) The hum continuously would sometimes be too much but did provide a level of privacy. (P6) This may be because SSS is artificial compared to NS. It can also be suggested that although a masking benefit may be had, it possible creates a feeling of tedium reflected in the larger change in ‘Interest and Understanding’. It was commented that SSI aided contextualisation of the soundscape with few other comments suggesting the soundscape was “no different” (P9): Gives better understanding of some unexpected sounds. (P5) [I] can contextualise the sounds more and felt more comfortable knowing what it was. (P18) More comfortable and relaxed. Not knowing the source of the noise is disconcerting. (P1) More accepting of noises I haven’t come across before. (P19) 5. Discussion The study successfully revealed that the soundscape interventions were effective in altering the emotional-cognitive response with a small but significant effect seen on the
Table 3 Difference and change in emotional-cognitive response caused by each condition. Emotional-cognitive response Relaxation Natural Sound Steady State Sound Sound Source Information Interest and Understanding Natural Sound Steady State Sound Sound Source Information
Mean difference (to control condition)
P-value
CI 95% lower bound e upper bound
Percentage change (control vs. intervention)
0.445 208 0.247
0.001 0.008 0.001
0.249e0.637 0.038e0.378 0.089e0.406
10.1% 3.3% 4.7%
0.00 -0.123 -0.066
1.00 0.338 1.00
0.6% 4.8% 2.0%
Please cite this article in press as: Mackrill, J., et al., Exploring positive hospital ward soundscape interventions, Applied Ergonomics (2014), http://dx.doi.org/10.1016/j.apergo.2014.04.005
J. Mackrill et al. / Applied Ergonomics xxx (2014) 1e7
‘Relaxation’ dimension. It was hypothesised that the soundscape would elicit more positive responses when each intervention was applied to soundscape clips. This was supported on the ‘Relaxation’ dimension with a less clear result was apparent on the ‘Interest and Understanding’ dimension. However, only a small effect size that was reported (partial h2 ¼ 0.05) across the main effect. Encouragingly, the scales reported good reliability (Cronbach’s a ¼ 0.895) across conditions and the measured emotional-cognitive response was reasonably robust. As it is necessary for sound within hospital environments to be considered beyond that of sound level (Blomkvist et al., 2005) discussion can be formed as to the potential of these soundscape interventions at a theoretical level. NS had a significant 10.1% effect on perception of the soundscape (p ¼
Contents lists available at ScienceDirect
Applied Ergonomics journal homepage: www.elsevier.com/locate/apergo
Exploring positive hospital ward soundscape interventions J. Mackrill*, P. Jennings, R. Cain WMG, University of Warwick, Coventry CV4 7AL, UK
a r t i c l e i n f o
a b s t r a c t
Article history: Received 30 July 2013 Accepted 6 April 2014 Available online xxx
Sound is often considered as a negative aspect of an environment that needs mitigating, particularly in hospitals. It is worthwhile however, to consider how subjective responses to hospital sounds can be made more positive. The authors identified natural sound, steady state sound and written sound source information as having the potential to do this. Listening evaluations were conducted with 24 participants who rated their emotional (Relaxation) and cognitive (Interest and Understanding) response to a variety of hospital ward soundscape clips across these three interventions. A repeated measures ANOVA revealed that the ‘Relaxation’ response was significantly affected (n2 ¼ 0.05, p ¼ 0.001) by the interventions with natural sound producing a 10.1% more positive response. Most interestingly, written sound source information produced a 4.7% positive change in response. The authors conclude that exploring different ways to improve the sounds of a hospital offers subjective benefits that move beyond sound level reduction. This is an area for future work to focus upon in an effort to achieve more positively experienced hospital soundscapes and environments. Ó 2014 Elsevier Ltd and The Ergonomics Society. All rights reserved.
Keywords: Soundscape Healthcare Environment
1. Introduction It is commonly accepted that the hospital environment affects patients’ experiences when they are admitted and treated in these spaces (see Dijkstra et al., 2006). Research on healthcare design and planning has highlighted strong relationships between environmental characteristics and human health (Monti et al., 2012). Sound is one part of the hospital environment. Ergonomics evaluations commonly consider the sound level of spaces, such as offices (Passero and Zannin, 2012) and consider the effect of them in reference to guidelines. Many studies have used this approach for hospitals and documented sound level within acute care hospital spaces (Busch-Vishniac et al., 2005; Okcu et al., 2011; Salandina et al., 2011). Sound levels range from 60 to 70 dB(A) with peaks of around 90 dB(A), equivalent to a busy motorway (Joseph and Ulrich, 2007). These exceed recommendations produced by bodies such as the World Health Organisation (Berglund et al., 2000) or the UK Government (Department of Health, 2008) by around 20e30 dB(A). Consideration is needed however, of how sound contributes to the feeling of the hospital space as ‘quietness’ is a quality indicator which is associated with a lack of annoying noises rather than the * Corresponding author. Tel.: þ44 24761 50760. E-mail address: [email protected] (J. Mackrill).
absence of sound itself (Fornara et al., 2006). Although there can be negative health consequences of excessive sound within these spaces (Ulrich, 1992), there is potential for positive benefits in understanding and maintaining the soundscape of the ward; for instance hearing the various occupational sounds, such as tea trolleys, as positive sounds (Mackrill et al., 2013a). A soundscape is defined as the auditory version of a landscape (Schafer, 1976) and considers the complete sound environment rather than focusing on a single source such as the beep of a monitor. As control of hospital sounds should consider positive aspects and move away from sound level measurement (Dawson, 2005) understanding the soundscape offers a new approach, particularly as the sound quality is not simply determined by level but other factors such as physical space (Hignett and Lu, 2010). Investigating the subjective response to the existing soundscape may create a more feasible way of improving the sound of the spaces, particularly as there is a consistent trend of rising levels since the 1960s (Busch-Vishniac et al., 2005). Therefore, a need to evaluate the effectiveness of interventions aiming to improve the subjective response to a hospital ward soundscape has arisen. Although studies have investigated subjective responses to hospital sounds to a certain degree (Topf, 1985; Akansel and Kaymakci, 2008; Xie and Kang, 2010) none have considered how to create a positive soundscape using interventions which may be of practical benefit.
http://dx.doi.org/10.1016/j.apergo.2014.04.005 0003-6870/Ó 2014 Elsevier Ltd and The Ergonomics Society. All rights reserved.
Please cite this article in press as: Mackrill, J., et al., Exploring positive hospital ward soundscape interventions, Applied Ergonomics (2014), http://dx.doi.org/10.1016/j.apergo.2014.04.005
2
J. Mackrill et al. / Applied Ergonomics xxx (2014) 1e7
1.1. Aim
Table 1 Sound clips and broad content classification for the 12 soundscape clip stimuli.
The aim of this study was to use laboratory listening evaluations to test interventions which might create a more positive emotionalcognitive response to hearing the soundscape of a cardiothoracic (CT) ward. This evaluated the effect of soundscape interventions to understand which may be considered appropriate for investigation within the ward setting. This was the third stage in a four part research project aiming at understanding and improving the perception of a hospital ward soundscape (Fig. 1). Part 1 aimed to understand the perception of the soundscape and inform the direction of subsequent steps. Part 2 created a framework to measure the response to the soundscape with this being used to understand the effect of soundscape interventions in Part 3, presented here. From this theoretical generalisation Part 4 was carried out in-situ to evaluate the chosen intervention. 2. Soundscape intervention rationale The following soundscape interventions were used based on the premise that sound level reduction would not necessary create a more positive soundscape. As a result it was hypothesised that the soundscape would elicit more positive responses when each intervention was applied to the existing sound clips of the CT soundscape. 2.1. Natural sound Natural sound (NS) was selected due to the agreement from literature (Guastavino, 2006; Pheasant et al., 2010; Yang and Kang, 2005a,b) which advocates that natural sounds of birdsong and water are positive soundscape features. Additionally, Mackrill et al. (2013a) found natural sounds to be a positive aspect of the CT ward soundscape when heard through windows overlooking a green space. 2.2. Steady state sound Steady state sound (SSS) was chosen as an intervention as masking sounds are a way in which negative sound can be controlled (Bowman, 1974). Defined as the presence of one sound that renders another sound undetectable (Plack, 2005), this has been tested to improve hospital soundscape perception (see Stanchina et al., 2005). When white noise was added to the sounds of an intensive care unit, despite an increase in sound level, subjectively sleep was consolidated and arousal was less frequent (Stanchina et al., 2005). Despite a small sample and a nonsignificant effect, this shows a change in perception.
Clip number
Playback dB(A)95%
Broad classification of dominant sounds
1 2
72.31 65.78
3
63.24
4 5 6
64.67 68.20 68.03
7
63.78
8
67.56
9
62.34
10
65.43
11 12
69.64 68.75
Conversation and sterilising machine sounds Conversation and footsteps. Doors opening and closing with objects banging. Conversation and footsteps in background. Trolley passing with objects banging, doors opening and closing. Quiet corridor. Objects banging in background. Deep rumble of trolley passing. Conversation and footsteps in background. Patient screaming intermittently. Conversation and footsteps in background. Trolley passing. Talking by nurse talking patient observations monitors beeping and sounds of a car from outside. Quiet patient bay with private conversation from nurse to patient in background. Quiet patient bay with monitor beeps, rustling, and talking in background. Floor cleaning buffer, long monitor beeps. Loud nurse conversation with laughter, monitor beeps.
2.3. Sound source information Written information detailing the various sound sources of the soundscape was chosen as hospital wards produce a sensory overload of unfamiliar stimuli which bombard patients with experiences they are unaccustomed to (Akansel and Kaymakci, 2008). Understanding hospital sounds however, enables patients to habituate and accept sound (Mackrill et al., 2013a). As complex environmental sounds are processed as meaningful events, providing information about interactions with environment (Guastavino et al., 2005), it may be necessary for patients to understand the sounds of the ward in order to feel comfortable within it. Furthermore, Topf (2000) suggests that personal control is the capacity to regulate stress caused by a negative event. This may be a cognitive approach by having information about the stressor (Topf, 2000). As such, it was proposed that providing information detailing what can be heard in the soundscape may improve the feeling that the soundscape elicits. This is called sound source information (SSI). 3. Method 3.1. Sound laboratory and stimuli As there is a challenge for clinicians, designers and researchers to work together using robust high quality research methods (Hignett and Lu, 2010) a laboratory listening evaluation method
Fig. 1. Timeline of the research project.
Please cite this article in press as: Mackrill, J., et al., Exploring positive hospital ward soundscape interventions, Applied Ergonomics (2014), http://dx.doi.org/10.1016/j.apergo.2014.04.005
J. Mackrill et al. / Applied Ergonomics xxx (2014) 1e7
was chosen to scope the potential of these interventions. This provided the means to facilitate the investigation of the inventions in an easily repeatable and controlled setting. Evaluations were held in a Sound Room laboratory. This consisted of a 16 speaker system (using KEF iQ70 and HTS3001SE speakers) operated via a PC computer using Nuendo 4 sound editing software. The system played back hospital soundscape recordings through all 16 speakers. Twelve twenty second soundscape excerpts were presented to participants. These were selected from a database of 32 recordings taken within the CT ward using a Bruel and Kjaer binaural SonoScout recording device. The most salient clips were selected for presentation informed by the analysis of all 32 recordings using the coding procedure developed by Poxon et al. (2009) (Table 1). For interventions of NS and SSS, the associated sounds were incorporated into all clips. NS was obtained from the website www. freesoundproject.org [accessed 3rd March 2011] and contained the sound of a blackbird singing and a babbling stream. SSS was obtained from using a short one second recording of a steriliser machine sound present within the CT ward providing a steady nondescript sound. The sound level of these interventions was set so they formed a congruent background feature rather than a dominant foreground sound. These sounds were only played through the ceiling speakers to help achieve this with the researcher judging the sound level at which they were set. SSI was presented on a response sheet (Table 2). This information was designed from the sound sources within the 12 soundscape clips. The information was the same for each soundscape clip rather than specifying details of the individual sources within each of the 12 clips. This was to assess if SSI would work in a generic manner. Further, as the hospital soundscape is constantly varying an in-situ SSI intervention would have to be generic in nature to accommodate the majority of features rather than focusing on specific nuances. 3.2. Experimental procedure Each soundscape clip lasted around 20 s presented sequentially with a 10 s inter stimulus period to discriminate between clips. Participants completed each of the four conditions on separate occasions. Interventions were applied to the 12 soundscape clips during each condition. Evaluations were conducted a minimum of two day apart to avoid demand effects, however this varied according to participant availability. To avoid order effects, participants were split into four groups with the condition sequence order randomised using a Latin square counterbalancing method. In addition, each group started the evaluation on a different soundscape clip, either 1, 2, 3 or 4, to further control for bias. Responses were recorded on a paper questionnaire using four seven point semantic differential scales. These rating scales were derived from the emotional-cognitive dimensions which the authors established in study 2 (see Mackrill et al., 2013b). The perceptual dimensions were labelled ‘Relaxation’ and ‘Interest and Understanding’. ‘Relaxation’ was measured using the bipolar semantic scales ‘relaxed-stressed’ and ‘comfortable-uncomfortable’. ‘Interest and Understanding’ used the scales ‘curious-uninterested’ and ‘intrigued-bored’. At the end of the questionnaire, basic demographic questions and open questions were asked to provide descriptive data beyond numerical quantification and included: Were there any sounds that particularly affected you? How did you feel overall when listening to the recordings? How does having information on the soundscape make you feel?
3
Table 2 SSI as an intervention used to detail the soundscape clip. What you may hear
Associated activities
Blood pressure monitors beeping for observations. Nurses talking to patients about how they feel. Trolleys moving equipment around the ward. Cleaning machines to keep the ward tidy. General bustling of the ward, patients going for treatments, phones ringing etc. Sound of trolleys bringing in food. Jingling of cups. Patients talking and moving around. Other staff talking.
Patient Observations. Cleaning. Bed Changing. Chatting. Washing.
Lunch.
3.3. Analysis The reliability of the scales measuring each dimension was tested across conditions to ensure that the results could be justifiably compared. This was carried out using a Cronbach’s a test. The main effect was analysed using a repeated measures analysis of variance (ANOVA) and was performed using statistical software SPSS 19 once the normality of the data had been established. 3.4. Sample Sample size was calculated using GPower3 software (Faul et al., 2007). Setting a power level of 0.80 with an effect size (eta2) of 0.25 (small) for a repeated measure ANOVA with four conditions and one group, sample size was calculated to be n ¼ 24 (a level set 0.05 resulting in a test power of 0.81, CI 95%). The 24 participants had a mean age of 32 years (S.D. 10.13 yrs), with 13 male and 11 female. The sample included a range of non-expert demographics from within and outside the University community. Recruitment took place using a convenience sampling strategy. Although a sample demographic was obtained that was removed from those specific to the CT ward, the soundscape of a CT ward was chosen due to the variety of sound representative of general hospital spaces. Therefore the sample demographic may experience such soundscapes which supported this strategy. Further this study was used to inform theoretical rather than empirical generalisation: generation of theoretical concepts deemed to be of wider application (Ritchie and Lewis, 2003). The exploration of empirical generalisation was carried out in stage 4 of the research project. 4. Results1 4.1. Scale reliability The ‘Relaxation’ dimension recorded high reliability, Cronbach’s
a ¼ 0.921 across all conditions. Likewise ‘Interest and Understanding’ recorded Cronbach’s a ¼ 0.895, suggesting that the scales were consistent in measuring the emotional-cognitive response and thus allowing a valid comparison between conditions. 4.2. Repeated measures ANOVA main effect The results showed a significant difference across all conditions on the ‘Relaxation’ dimension; (F (3,751) ¼ 13.991, p ¼ 0.001). A small overall effect size of 0.05 (partial h2) showed that 5% of the variation in emotional-cognitive response can be accounted for by
1 All analysis was performed on the combined mean scores for each dimension i.e. ‘Relaxation’ was measured using the mean scores of scales stressed-relaxed and comfortable-uncomfortable; ‘Interest and Understanding’ using the mean score of scales curious-uninterested and intrigued-bored.
Please cite this article in press as: Mackrill, J., et al., Exploring positive hospital ward soundscape interventions, Applied Ergonomics (2014), http://dx.doi.org/10.1016/j.apergo.2014.04.005
4
J. Mackrill et al. / Applied Ergonomics xxx (2014) 1e7
Private conversation between patients and nurses I don’t feel I should be listening to these. (P12) The crying lady [sound clip 6, a patient with dementia crying out] was upsetting. (P3) [I] loath beeping noise. Grates a bit. (P13) Mixture of boredom and curiosity to generally unpleasant soundscape. (P24) Fluid bags didn’t sound nice when rustling. (P5)
Control
7
Natural Sound Steady State Sound 6
Sound Source Information
5 Relaxation 1
2
4
3
4
5
6
7
The addition of NS was described as: 3
Relaxing. (P3) I find hospital make me feel claustrophobic. The birdsong and water make it feel less so like being next to a window. (P19) Bought a soothing dimension. (P24)
2
1 Interest & Understanding
However, despite positive comments some were negative:
Fig. 2. The effect of each intervention on each clip mapped within the twodimensional space using the mean scores for each condition. Note, increase in ‘Relaxation’ in interventions with the highest NS and SSI.
the differing conditions. A non-significant difference was seen between conditions on the ‘Interest and Understanding’ dimension (F(3,764) ¼ 1.447, p ¼ 0.229) (Fig. 2). Pairwise comparisons, with a Bonferroni correction to account for the increase in pairwise comparisons, showed a significant difference between the control and all interventions (Table 3). NS caused the largest change in response; 10.1%; (mean difference NS ¼ 0.445, p ¼ 0.001), CI (95%) 0.249e0.637). SSS had a smaller difference causing a 3.3% change; (mean difference SSS ¼ 0.208, p ¼ 0.008. CI (95%) 0.038e0.378). Finally, SSI was successful in producing a moderate difference in scores almost half that of natural sound with 4.7%; (mean difference SSI ¼ 0.247, p ¼ 0.001), CI (95%) 0.089e0.406). Pairwise comparisons for ‘Interest and Understanding’ explored the smaller difference between the control condition and interventions (Table 3). The mean difference of NS ¼ 0.00, p ¼ 1.00 corresponding to a 0.6% change. However, SSS caused a larger nonsignificant change of 4.7%, mean difference SSS ¼ 0.123, p ¼ 0.338. Finally, SSI produce a small effect mean difference ¼ 0.066, p ¼ 1.00, resulting in a 2% change. 4.3. Subjective comments Although the manner of each intervention was not divulged to participants, subjective response to the conditions showed trends in response to positive and negative sounds. Negative sounds included patient and staff conversation, particularly when private. Sounds of monitor alarms were also described as annoying:
I like the birdsong but it could get too much if ‘piped’ in. (P3) With birds and conversation I felt life was passing me by. (P18) Sound of running water didn’t fit. (P6) To be left in quiet all day would drive me just as mad as listening to alarms. (P8)
The result of SSS was considered more negative with no positive free responses obtained: White noise and beeps don’t make me relaxed. (P8) The hum continuously would sometimes be too much but did provide a level of privacy. (P6) This may be because SSS is artificial compared to NS. It can also be suggested that although a masking benefit may be had, it possible creates a feeling of tedium reflected in the larger change in ‘Interest and Understanding’. It was commented that SSI aided contextualisation of the soundscape with few other comments suggesting the soundscape was “no different” (P9): Gives better understanding of some unexpected sounds. (P5) [I] can contextualise the sounds more and felt more comfortable knowing what it was. (P18) More comfortable and relaxed. Not knowing the source of the noise is disconcerting. (P1) More accepting of noises I haven’t come across before. (P19) 5. Discussion The study successfully revealed that the soundscape interventions were effective in altering the emotional-cognitive response with a small but significant effect seen on the
Table 3 Difference and change in emotional-cognitive response caused by each condition. Emotional-cognitive response Relaxation Natural Sound Steady State Sound Sound Source Information Interest and Understanding Natural Sound Steady State Sound Sound Source Information
Mean difference (to control condition)
P-value
CI 95% lower bound e upper bound
Percentage change (control vs. intervention)
0.445 208 0.247
0.001 0.008 0.001
0.249e0.637 0.038e0.378 0.089e0.406
10.1% 3.3% 4.7%
0.00 -0.123 -0.066
1.00 0.338 1.00
0.6% 4.8% 2.0%
Please cite this article in press as: Mackrill, J., et al., Exploring positive hospital ward soundscape interventions, Applied Ergonomics (2014), http://dx.doi.org/10.1016/j.apergo.2014.04.005
J. Mackrill et al. / Applied Ergonomics xxx (2014) 1e7
‘Relaxation’ dimension. It was hypothesised that the soundscape would elicit more positive responses when each intervention was applied to soundscape clips. This was supported on the ‘Relaxation’ dimension with a less clear result was apparent on the ‘Interest and Understanding’ dimension. However, only a small effect size that was reported (partial h2 ¼ 0.05) across the main effect. Encouragingly, the scales reported good reliability (Cronbach’s a ¼ 0.895) across conditions and the measured emotional-cognitive response was reasonably robust. As it is necessary for sound within hospital environments to be considered beyond that of sound level (Blomkvist et al., 2005) discussion can be formed as to the potential of these soundscape interventions at a theoretical level. NS had a significant 10.1% effect on perception of the soundscape (p ¼
