LOW FREQUENCY NOISE DURING WORK - EFFECTS ON PERFORMANCE AND ANNOYANCE

JOHANNA BENGTSSON

Göteborg University – 2003

LOW FREQUENCY NOISE DURING WORK - EFFECTS ON PERFORMANCE AND ANNOYANCE

Johanna Bengtsson

Department of Environmental Medicine, The Sahlgrenska Academy, Göteborg University, Box 414, S-405 30 Göteborg, Sweden Telephone: +46 (0)31 773 36 21, Fax: +46 (0)31 82 50 04, E-mail: [email protected]

ISBN 91-628-5639-1

2

ABSTRACT Johanna Bengtsson, Department of Environmental Medicine, The Sahlgrenska Academy, Göteborg University, S-405 30 Göteborg, Sweden. Thesis defended April 23, 2003. Aims. Low frequency noise (LFN) is defined as “a noise with a dominant frequency content of 20 to 200 Hz”. Common sources of LFN in occupational environments are ventilation, heating and air-conditioning systems, computer network installations and compressors. The aims were to evaluate the influence of LFN on performance, annoyance, other subjective effects, cortisol levels and subjective stress. A further aim was to evaluate whether the frequency balance and modulation frequency in a LFN influenced a subject’s perception of pleasantness. Methods. All studies were laboratory experiments. The experiment reported in papers I and II comprised 32 subjects who worked for 2 h with four performance tasks under high workload during exposure to LFN and a reference noise at an A-weighted sound pressure level of 40 dB. The experiment reported in paper III comprised 38 subjects who worked for 4 h with six performance tasks under low workload during exposure to LFN or reference noise at an Aweighted sound pressure level of 45 dB. The experiment reported in paper IV comprised 30 subjects who varied the level of the sound characteristic’s frequency balance and modulation frequency in the LFN to make the noise more pleasant. Results. LFN impaired performance in tasks with high and moderate demands on cognitive processing when carried out under high workload and in tasks with moderate and low demands when these were performed under low workload. No difference between noise conditions was found in low demand tasks performed under high workload and tasks evaluating motivation performed under low workload. LFN was rated to have a greater impairment on the work capacity and be more annoying than reference noise; the difference between noises was significant under high workload. No difference between noise conditions was found in subjective symptoms, but annoyance and reported impairment of the work capacity due to LFN was related to several symptoms. This was less frequently found for reference noise. Subjects high-sensitive to LFN or to noise in general performed less well and reported higher annoyance due to LFN. The effects caused by LFN were most pronounced for subjects high-sensitive to LFN. Exposure to LFN during high workload resulted in elevated cortisol levels among subjects high-sensitive to noise in general, and a tendency towards the same result was found for subjects high-sensitive to LFN. No clear relationships between cortisol levels and subjective stress were found. The resulting pleasant LFN comprised less perceivable modulations and a lower content of frequencies below 500 Hz. The effect was less marked when the original LFN did not comprise modulations and, when the A-weighted sound pressure level could be altered, a steeper slope was preferred to a higher level. Conclusions. The experiments showed that exposure to LFN during work can impair performance, lead to subjective annoyance and increase cortisol levels, even at moderate sound pressure levels. The effects were influenced by workload and noise sensitivity. The performance effects are hypothesised to be mediated by impaired learning and reduced attention. To achieve a more pleasant LFN, the noise should contain no or little perceivable modulations and a lower relative content of low frequencies. Key words: Low frequency noise, performance, workload, annoyance, cortisol, subjective stress, subjective sensitivity to noise, sound characteristics, frequency balance, modulation frequency.

3

The thesis is based on the following four papers, two of which are from the same study (paper I and II). The papers will be referred to in the text by their Roman numerals.

I:

Persson Waye K, Bengtsson J, Kjellberg A and Benton S. Low frequency noise “pollution” interferes with performance. Noise & Health, 2001;4: 33-49.

II:

Persson Waye K, Bengtsson J, Rylander R, Hucklebridge F, Evans P and Clow A. Low frequency noise enhances cortisol among noise sensitive subjects during work performance. Life Science, 2002;70: 745-758.

III:

Bengtsson J, Persson Waye K and Kjellberg A. Evaluations of effects due to low frequency noise in a low demanding work situation. J Sound Vib 2003 (Submitted).

IV:

Bengtsson J, Persson Waye K and Kjellberg A. Sound characteristics in low frequency noise and their relevance for the perception of pleasantness. Acta Acoustica 2003 (Submitted).

4

ERRATA 2004-11-01 Bengtsson J, Persson Waye K and Kjellberg A. Sound characteristics in low frequency noise and their relevance for the perception of pleasantness. Published in: Acta Acoustica 2004;90:171-180. Bengtsson J, Persson Waye K and Kjellberg A. Evaluations of effects due to low-frequency noise in a low demanding work situation. Published in: J of Sound and Vib 2004;278;1-2:83-99. ERRATA 2003-04-23 THESIS Page 11, 1st paragraph, line 1, ”In 1851…” change to ”In 1859…” Page 13, 3rd paragraph, line 5, ”…perceived loudness.” change to ” …perceived loudness level.” Page 14, 1st paragraph, line 4, ”…led to less annoyance” change to ” …led to increased annoyance” Page 28, 1st paragraph, line 9 and 11, ”Schultz” change to ”Schulz” Page 28, 2nd paragraph, line 2, ”Ising and Braun 2002” change to ”Ising and Braun 2000”. Page 35, 2nd paragraph, line 2, ”between-subject design.” change to ”within-subject design.” Page 72, 3rd paragraph, line 8, ”Furthermore” change to ”Therefore”. Page 75, 1st paragraph, line 2, ”e.g. dBA” change to ” i.e. dBA”. REFERENCES Page 91, “Schultz P, Merck D“... change to “Schulz P, Merck D…” Page 91, ”Schultz P, Kirschbaum C, Prüssner J, Hellhammer D…“ change to “Schulz P, Kirschbaum C, Prüssner J, Hellhammer D...“ Page 91, delete Smith AP, Broadbent DE. Noise and levels of processing. Acta Psychologica 1981;47:129-142. PAPER I Page 47, 2nd column, 2nd paragraph, line 15, ”categorisation of sensitivity to low frequency noise, while this difference was not found using the categorisation of sensitivity to noise in general” change to ”categorisation of sensitivity to noise in general, while this difference was not found using the categorisation of sensitivity to low frequency noise”.

5

6

CONTENTS

LIST OF ABBREVIATIONS AND EXPRESSIONS ..................................................................9 INTRODUCTION................................................................................................................... 11 BACKGROUND ...................................................................................................................... 13 Low frequency noise ................................................................................................... 13 Prevalence of people exposed to low frequency noise ........................................ 15 Prevalence in the general environment.......................................................... 15 Prevalence in the occupational environment................................................ 16 Effects of low frequency noise in occupational environments.......................... 17 Performance........................................................................................................... 17 Subjective effects .................................................................................................. 23 Symptoms.......................................................................................................... 23 Annoyance ........................................................................................................ 25 Stress induced by noise ....................................................................................... 26 Sensitivity to noise...................................................................................................... 30 Summary of this review............................................................................................. 31 AIMS ....................................................................................................................................... 32 METHODS .............................................................................................................................. 33 Scientific approach ..................................................................................................... 33 Study designs ............................................................................................................... 33 Exposure noises ........................................................................................................... 35 Performance tasks....................................................................................................... 37 Questionnaires............................................................................................................. 40 Saliva sampling and cortisol determination.......................................................... 41 Subjects ......................................................................................................................... 42 Subjective sensitivity to noise ........................................................................... 42 Test chamber ............................................................................................................... 44 Learning session .......................................................................................................... 45 Statistical treatment .................................................................................................. 45 MAIN RESULTS ..................................................................................................................... 47

7

Paper I ........................................................................................................................... 47 Paper II.......................................................................................................................... 50 Paper III ........................................................................................................................ 52 Paper IV......................................................................................................................... 55 DISCUSSION.......................................................................................................................... 57 Background................................................................................................................... 57 Methods......................................................................................................................... 57 Results ........................................................................................................................... 61 Effects of low frequency noise on performance............................................ 61 Effects of low frequency noise on annoyance and other subjective reports ..................................................................................................................... 64 Effects of low frequency noise on cortisol and subjective stress.............. 68 The influence of subjective sensitivity to noise for adverse effects ........ 72 Sound characteristics .......................................................................................... 73 Continuation of the work in this thesis - further studies.................................. 76 CONCLUSIONS ...................................................................................................................... 77 ACKNOWLEDGEMENTS ....................................................................................................... 78 REFERENCES.......................................................................................................................... 80

8

LIST OF ABBREVIATIONS AND EXPRESSIONS

Below is a list of terms used in this thesis and in the articles referred to in the text.

dBA

A- weighted sound pressure level (dB)

dBB

B- weighted sound pressure level (dB)

dBC

C- weighted sound pressure level (dB)

dB SPL

sound pressure level (dB)

dB Lin

linear sound pressure level (dB)

LAeqT

the equivalent A-weighted sound pressure level integrated during the time period (T)

LCeqT

the equivalent C-weighted sound pressure level integrated during the time period (T)

LAmax

the maximum momentary A-weighted sound pressure level, during the time period (T)

Hz

hertz

ISO

International Organization for Standardisation

ECG

electrocardiography

EEG

electroencephalography

Infrasound

sound in the frequency range below 20 Hz

White noise

noise that include all frequencies equally strong

Pink noise

noise including all frequencies but filtered so each octave band is equally strong

9

10

INTRODUCTION

Knowledge that noise may cause effects on humans goes back many years. In 1851, the legendary English nurse Florence Nightingale pointed out the importance of a good indoor environment at the hospital, writing in the textbook “Notes on Nursing” that ”unnecessary noise, then, is the most cruel absence of care which can be inflicted either on sick or well” [Nightingale 1980]. Since then, a vast amount of research has been done on noise and its effect on people. However, only a small portion of this research has been done on noise dominated by low frequencies. This thesis deals with low frequency noise in working environments such as office areas and control rooms. There is no international agreement on how exactly to define low frequency noise. The upper limit according to different authors varies between 100 and 500 Hz while the lower limit usually is set at 20 Hz. In this thesis, low frequency noise is defined as “a noise with a dominant frequency content of 20 to 200 Hz” [Persson Waye 1995]. From an acoustical point of view, there is no difference between the expressions sound and noise. The most common definition of noise is unwanted sound [Berglund et al. 1995]. Noise can be defined as being “received” (attributes of noise annoyance), in contrast to sound, which is “perceived” (attributes of sound sensation) [Preis 1996]. The common way to describe a noise is to make physical measurements but, according to these definitions, there is in reality no other descriptor than a person’s subjective perception of a sound that is of relevance for when sound should be defined as noise. Low frequency sound is an important tool for communication between animals, and tigers, whales, elephants, rhinos and other animals use this [Von Muggenthaler 2003]. Low frequency sounds can also be considered an important ingredient in music that strengthens the overall impression. Not all low frequency sounds are experienced as pleasant and/or necessary, however. Much of the early work on infrasound started with the purpose of evaluating possible adverse effects on astronauts during their space trips. Measurements today aboard the International Space Station have shown that the acoustic environment of the astronaut and cosmonaut crews is dominated by low frequencies caused by noise from life

11

support systems and experiments that use fans, pumps and motors [Brüel & Kjær 2000]. Effects of low frequency noise in the general environment have been explored less than noises of higher frequencies. The importance of low frequency noise was pointed out in a document on community noise prepared for the WHO [Berglund et al. 2000] that stated that noise that includes a large proportion of low frequency components calls for lower values than the existing guideline values for community noise. Several studies show increasing evidence that exposure to low frequency noise may cause effects that are different than those of other environmental noises of comparable levels but which are not dominated by low frequency components [Persson Waye 1995; Berglund et al. 1994]. Tiredness, headaches and irritation are symptoms that have been found in connection with annoyance caused by low frequency noise [Tokita 1980; Nagai et al. 1989; Persson Waye and Rylander 2001]. It is possible that the presence of such symptoms could lead to a poorer work capacity. The sensation of a low frequency noise can also, at sound pressure levels above the hearing threshold, be perceived as pressure over the eardrums or head, or, at higher levels, as vibrations in the chest and the stomach [Landström et al. 1999]. Common sources of low frequency noise in occupational environments are ventilation, heating and air-conditioning systems, computer network installations and compressors. Little is known about how people exposed to low frequency noise during work in occupational environments are affected. On the basis of existing knowledge of effects of low frequency noise in the general environment, however, it can be hypothesised that exposure to low frequency noise during work may cause subjective effects and/or negatively affect work capacity. Effects of low frequency noise on people in the general environment have previously been described in the thesis “On the effects of environmental low frequency noise” [Persson Waye 1995], which evaluated the proportions of problems caused by exposure to low frequency noise in the general environment. In the present thesis, the focus is on the occupational environment and the aim is to study effects on performance, annoyance and other subjective effects and stress that are caused by exposure to low frequency noise during work. The scope of the thesis is limited to exposure to steady-state low frequency noise at levels commonly occurring in two different kinds of occupational working environments, office areas and control rooms, through a series of experimental laboratory studies. Effects of low frequency noise are studied during work with tasks that reflect real assignments or assignments reflecting relevant demands in the two different working environments. 12

BACKGROUND

Low frequency noise

A common way of describing a noise is in A-weighted sound pressure levels. While Aweighted sound pressure levels correspond relatively well with effects such as hearing impairment and annoyance resulting from e.g. transportation noise, several studies have indicated that an A-weighted noise is a less suitable descriptor for assessing effects of low frequency noise [see e.g. Kjellberg and Goldstein 1985; Persson et al. 1990; Persson Waye 1995; Persson Waye and Rylander 2001] and that measuring the A-weighted sound pressure level is not a sufficient way of characterising a noise environment dominated by low frequencies [Berglund et al. 2000]. In 1996, Sweden adopted a specific guideline for low frequency noise in the general environment [SOSFS 1996:7]. This guideline comprises specific recommendations for low frequency noise that are based on third-octave band measurements in the frequency range of 31.5 to 200 Hz. However, the present guideline for occupational environments is described in A-weighted sound pressure levels or for infrasound, in third-octave band sound pressure levels [AFS 1992:10]. No specific recommendations for the occupational environment exist that comprise particular descriptions of how to assess low frequency noise, and there is therefore a need to develop alternative measures that can better predict negative effects of low frequency noise. If an A-weighted sound pressure level is an unsuitable descriptor of low frequency noise, further knowledge is needed of how to correctly describe low frequency noise in order to avoid negative effects on annoyance and work performance. One important factor in comparing low frequency noise to other noises not dominated by low frequencies is the presence of different sound characteristics. One sound characteristic is perceived loudness. An increase of as little as 5-6 dB in low frequencies is perceived as a two-fold increase in the subjective loudness, while 10 dB is required for the same change in sensation for higher frequencies [ISO 226, 1987].

13

Bryan [1976] suggested that the slope and the turnover point (a breakpoint at which the slope does not continue to rise) were more important for annoyance than the sound pressure level. He meant that a decrease in the slope would result in decreased annoyance. The opposite was found by Goldstein and Kjellberg [1985], who suggested that a decreasing slope led to less annoyance, while Holmberg et al. [1993] did not find a relation between annoyance caused by ventilation noises with gradually falling frequency and two other ventilation noises at the same A-weighted sound pressure level. The difference in the results demonstrates that further investigations of the slope theory are necessary to understand the influence of slope for perceived annoyance. Noises with a dominating content of low frequencies often comprise modulations that give them a rumbling character [Broner 1994]. A few studies have investigated the importance of modulations in low frequency noise. Persson Waye et al. [1993] found that a sinusoidal tone at 31.5 Hz, modulated with 2.5 Hz, led to higher degrees of drowsiness than a non-modulated sinusoidal tone at the same frequency. In another study, Landström et al. [1996] compared two modulated tones (at 100 Hz and 1000 Hz) and two modulated broadband noises (with centre frequencies at 100 Hz and 1000 Hz), all at 50 dB LAeq. The noises were modulated with 0.1 Hz, 0.2 Hz, 0.4 Hz, 0.8 Hz, 1.6 Hz, 3.2 Hz, 6.4 Hz, 12.8 Hz, 25.6 Hz and 51.2 Hz. According to the results, a sound modulated with 2-3 Hz was rated as more annoying than both lower and higher modulation frequencies. Furthermore, Zwicker and Fastl [1999] found that a modulation frequency of 4 Hz was most clearly perceived. In comparison to the presence of modulations in low frequency noise, the influence of a tonal character in the low frequency range has been shown to be of little or no importance for annoyance, reduced wakefulness or performance [Holmberg et al. 1993; Landström et al. 1991; Landström et al. 1995]. In summary, an A-weighted sound pressure level has been found to be an unsatisfactory descriptor to predict effects caused by low frequency noise. According to previous studies, specific sound characteristics that are not fully assessed by an A-weighted sound pressure level could be of importance for adverse effects of low frequency noise. This thesis focuses on the sound characteristics of frequency balance (i.e. the relative content of high and low frequencies) and modulation frequency.

14

Prevalence of people exposed to low frequency noise

Prevalence in the general environment

The number of people exposed to transportation noise can rather easily be estimated using standardised calculations, e.g. the Nordic Prediction Method [Kragh et al. 1996], based on number of vehicles, vehicles´ average speed, the distance from the road to the dwellings etc. For low frequency noise, it is more difficult to estimate how many are exposed. There is no information on the number of low frequency sources to which people may be exposed. Furthermore, some usually irrelevant sources may generate a low frequency noise because the installations are inaccurate. One way to estimate the number of people exposed to low frequency noise is to use indirect estimations, such as the prevalence of complaints. However, it is important to remember that the reasons for making complaints depend on several factors related to the individual, the noise source and the society. Hence the number of complaints does not automatically reflect how many are actually disturbed but can be used as one indicator of disturbance. Mirowska [1998] conducted a study among persons who complained about noise in their homes, where the noise levels, according to measurements made by the Local Health Authorities, did not exceed the recommended A-weighted sound pressure level. It was found that noises from installations heard at home were more annoying than traffic noise. Furthermore, higher levels of traffic noise seemed to be more acceptable than the noise from the installations [Mirowska 1998]. Two studies in Sweden have evaluated how many complaints about noise in general and how many complaints about low frequency noise in particular that were received by the local Environmental Health Authorities (EHA). Persson and Rylander [1988] carried out a questionnaire study among all local EHAs, and the results showed that three low frequency noise sources, fans and ventilation, heavy vehicles and heat pumps, accounted for 71% of the total number of complaints about noise during a six-month period in 1985. Fourteen years later, Bengtsson and Persson Waye [2003] conducted an interview study on a selection (13%) of the local EHAs. Complaints about low frequency noise comprised 35% of the complaints on noise that had been received. Most of the complaints named noise from fans and ventilation, compressors, music and laundry rooms. 15

Three factors could explain the difference in the results in a comparison of the two studies. The relative increase in the number of complaints on noise in general in the later study, an increased knowledge about low frequency noise and that better procedures now exist for measurements that can more precisely detect a low frequency noise. Considering this, it is not unlikely that some of the complaints in the previous study were incorrectly classified as having been prompted by low frequency noise.

Prevalence in the occupational environment

Interest in noise as intruding in the occupational environment has become broader, from earlier focusing on high levels of noise that can cause hearing impairments. Today, research is giving increasing attention to non-auditory effects caused by noise, e.g. effects on work performance, annoyance and subjectively judged health. People working in office areas and control room environments are exposed to a large number of sources of noise at moderate sound pressure levels, such as speech, computers, computer network installations and ventilation systems. The most common sources of low frequency noise in working environments are ventilation, heating and air-conditioning systems, compressors and diesel engines [Landström et al. 1999]. In a study among 30 air-conditioned open-plan offices, it was found that one-third of the 1080 employees reported the air-conditioning systems as the major noise source, followed by noise from e.g. telephones and humans [Tang et al. 1996]. Tang et al. [1996] concludes that the air-conditioning systems were perceived as the most dominant noise sources when the noises were rumbling and had a rather great low frequency imbalance of the spectrum. In another study, ventilation noises containing higher levels of low frequency components were correlated to higher levels of annoyance [Landström et al. 1991]. Low frequency noise is more difficult to attenuate than noises that are not dominated by low frequencies because of the relatively lower reduction of the low frequencies by walls, floors and ceilings. As a result, a noise with a mixed content of frequencies will be selectively attenuated, and the resulting sound will normally be dominated by low frequencies. An example of this problem is achieving a reduction in high noise exposure levels by building control rooms from which industrial processes can be supervised. A new noise problem may

16

be that the work environment in the control room now comprises noise dominated by low frequencies. In summary, it is not possible to estimate how many people are exposed to low frequency noise in their working environment today. Low frequency noise may be a problem at work places with large ventilation systems or when sounds with a mixed frequency content are attenuated and, the attenuated noise may be a noise dominated by low frequencies.

Effects of low frequency noise in occupational environments

The following chapters review the major adverse effects of exposure to low frequency noise that are of relevance for this thesis. The Swedish guideline “Ordinance on noise” [AFS 1992:10] recommends that 40 dB LAeq8h should not be exceeded for working conditions that make great demands on concentration and undisturbed conversation, tasks relevant to e.g. office areas. Furthermore, 60 dB LAeq8h should not be exceeded for working conditions with strict demands on attention, precision and speed and on undisturbed conversation, tasks that are relevant in e.g. control rooms. The focus on present knowledge in this thesis is primarily on possible adverse effects within these A-weighted sound pressure levels and these types of working environments.

Performance

Over the years, a fair amount of research has been done to evaluate adverse effects on performance from different kind of noises [for reviews see e.g. Smith and Jones 1992; Kjellberg and Landström 1994]. Many of these studies have been based on noise at rather high levels. There is less information on what influence noise of more moderate sound pressure level can have on performance and even less data exist on moderate levels of low frequency noise. A number of studies are of particular interest for this thesis. For example Benton and

17

Leventhall [1986] compared pure tones centred at 40 Hz and 100 Hz (both modulated at 1 Hz) and a narrow band noise centred at 70 Hz, all at a level of 25 dB above the individual hearing threshold, and recorded traffic noise (90 dB Lin) and a silent control condition. They found that the tones centred at 40 Hz and 100 Hz caused more errors in a dual task situation, i.e. when the subjects worked with two tasks at the same time, compared with the traffic noise and silence. The effects were especially pronounced during the last ten minutes of the total 30-minute exposure. Further support for performance impairment caused by low frequency noise is given in Benton and Robinson [1993], where it was found that under conditions of narrow band low frequency noise at 70 dBC or 95 dBC, subjects made more semantic and spelling errors on a proofreading task. The subjects also rated the low frequency noise as more annoying than two other noise conditions (speech and white noise, 20-20k Hz) matched for loudness against the narrow band low frequency noise. Kjellberg and Wide [1988] found a lower learning rate in a demanding verbal grammatical reasoning task when the task was performed during exposure to simulated ventilation broadband noise (15-1000 Hz, 51 dBA, 57 dBD) with a dominance of energy in the low frequency area. The comparison was made between two groups working with the task for 25 minutes. In the first group, the noise was turned on after five minutes and, in the second group, the noise was turned on after 20 minutes. Similar results using the same task were reported by Persson Waye et al. [1997]. During exposure to a low frequency ventilation noise (A-weighted sound pressure level 42 dB), there was a tendency among the subjects to require a longer time to respond in the task as compared to working during exposure to a ventilation noise of equal A-weighted sound pressure level but not dominated by low frequencies. However, other studies have not found a difference when comparing low frequency noise with other noises not dominated by low frequency components. Key and Payne [1981] found no effects in subjects working on a complex task during exposure to high levels of a low frequency noise (one-third-octave band centred at 125 Hz, 90 dB SPL) as compared to a high frequency noise (linear sound pressure 90 dB) or ambient noise (linear sound pressure 55 dB). Several factors have been found to be important for effects on performance. Most of the documented effects are based on exposure to noise at rather high levels. Kjellberg and Landström [1994] summarized in their review that tasks of more complex character could be especially negatively affected when performed during exposure to noise, even at moderate sound pressure levels. Furthermore, it is suggested that noise exposure can make a task more demanding, and thereby also impair performance. This means that how well a task can be 18

performed in noise depends not only on the task but also on the working conditions, i.e. the workload under which the task is performed. The workload is set by task demands and the pressure put upon a person. By varying these factors, different aspects on performance can be evaluated. On the basis of the knowledge of effects caused by low frequency noise, it could be hypothesised that low frequency noise may lead to specific performance impairments and that exposure to low frequency noise during work may be more or less serious depending on workload. To draw conclusions about what critical task demands that are sensitive to low frequency noise, the choice of performance tasks would have to be based on hypotheses about possible mechanisms behind such effects. Previous studies suggest that low frequency noise may be more difficult to ignore or habituate to and that it can have a sleep-provoking effect and thus lead to increased tiredness. With better knowledge of the mechanisms involved, there will be greater opportunity to create a better working environment. In a discussion of theories and mechanisms that may explain effects on performance, two theories are of major interest – the arousal theory and the information overload theory. Brief descriptions of these two theories are given here [for greater detail, see e.g. Hebb 1955; Broadhurst 1959; Broadbent 1971; Kahneman 1973; Hockey 1979; Cohen 1980; Cohen et al. 1986; Jones and Davies 1984; Kjellberg and Landström 1994]. The arousal theory describes the balance between arousal level and performance. Starting at a very low level of arousal, there is initially an increase in alertness, interest, and positive emotion up to a level at which the response and learning is optimal. When the optimal level is exceeded, there is instead an increase in emotional disturbance and anxiety [Hebb 1955]. When the arousal level is too low, negative effects on performance can be explained by fatigue, while, when the arousal level is too high, negative effects may be caused by a narrowing of attention and focus [Cohen et al. 1986]. Figure 1 shows the Yerkes-Dodson model of the relationship between arousal and performance [redrafted from Kahneman 1973]. Kahneman [1973] explained the original findings by Yerkes-Dodson [1908] such that “the quality of performance on any task is an inverted U-shaped function of arousal, and that the range over which performance improves with increasing arousal varies with task complexity”. This means that the optimal level of arousal is different for different kinds of tasks. The optimal level to successfully complete a simple task is higher than a lower level to successfully complete a more complex task 19

[Kahneman 1973]. Thus, when the arousal level is above or below a task’s optimal level, performance goes down.

High

Quality of performance

A simple task

A complex task

Low Low

Arousal level

High

Figure 1. The Yerkes-Dodson model of arousal and performance [redrafted from Kahneman 1973]. As will be hypothesised below in the description of the relationship between tiredness and motivation, tiredness may reduce motivation and greater motivation may influence tiredness. Subjects with a low arousal level may perform less well due to poor attention and may also have less motivation to perform well. Monotonous, machine-paced tasks are most sensitive to changes in wakefulness [Hockey 1986]. Previous studies suggest that low frequency noise can have a sleep-provoking effect, which means that exposure to low frequency noise can result in reduced wakefulness [e.g. Landström et al. 1982; Landström et al. 1983; Landström 1987; Landström 1990; Kjellberg et al. 1998]. If a task is sensitive to a lower level of wakefulness, i.e. increased tiredness, the task can also be particularly sensitive to exposure to low frequency noise. Examples of such tasks are monotonous, repetitive tasks or survey tasks, where a computer or a machine determines the working pace, i.e. tasks with demands on sustained attention. 20

The information overload theory was first proposed by Broadbent [1957]. The theory describes how information is given priority, i.e. how sources of information of lower priority are filtered out in order for a task to be performed in the best possible way. Cohen et al. [1986] concluded that a person’s cognitive capacity can be reduced when a stressor requires attention, and this may result in a lower capacity to e.g. complete a task. The most probable way to handle an overload of information may be to filter out those parts of the information that are considered to be of less importance. Broadbent [1957] tried to explain the perceptual system by comparing it to a Y-shaped tube, where imaginary balls are trying to enter the tube. These balls represent information from different kinds of stimuli, and the conclusion to be made from this metaphor is that the perceptual system has a limited capacity. “A feeling of relief” is commonly reported when low frequency noise is turned off, even if the person was not aware of the noise when it was present [Landström et al. 1991; Kjellberg and Wide 1988]. Empirical findings also suggest that a low frequency noise is more difficult to ignore or habituate to, as compared to other noises not dominated by low frequencies [Benton 1997]. Low frequency noise has also been suggested to act as a background stressor, and the result can either be direct intrusion by affecting attention or a displacement of cognitive and perceptual processes [Benton and Leventhall 1994]. These studies can be interpreted to mean that low frequency noise acts as some kind of background stressor that the central part of our auditory system tries to filter out in order for us to be unconsciously aware of the noise signals. This filtering process is however believed to be energy demanding and hence to influence our mental capacity. People annoyed by low frequency noise have described it in terms of “it feels all around”, “cannot be ignored”, is “worse indoors”, “cannot locate it” and also “tuned into it” [Benton 1997]. Furthermore, low frequency noise can be perceived as annoying and causing interference even though it is not experienced as very loud. If there is little or no habituation to low frequency noise, low frequency noise will interfere with and demand a subject’s attention. This can result in an impaired ability to process and perform mentally demanding tasks, due to a competition between available mental resources. As a result, the tasks may be additionally strenuous as the subject uses part of her mental capacity on the noise exposure. This may lead to lower performance capacity and/or quality. Furthermore, the results from two studies indicate that the effort to cope in low frequency noise during work develops over time and, it could therefore be more demanding over time to work during exposure to low frequency noise [Benton 1997; Persson Waye et al. 1997]. 21

To be able to cope with the extra load from e.g. noise during work, there is a need for subjects to change their way of working to perform the task equally well despite the extra load [Hockey 1979]. Several authors have described strategies to cope with extra load, adopted either consciously or unconsciously. Strategies that may be relevant for work during exposures to low frequency noise could be to work more rapidly and less thoroughly [Broadbent 1971; Kahneman 1973; Smith and Jones 1992], to work at a slower working speed or to make extra effort to complete the task correctly [Kahneman 1973; Landström et al. 1997a]. Another strategy could be to continue to work as before even though the working condition has changed. This can manifest itself as a dislike of e.g. learning and developing a better way of performing a task. One example of this was found in a study among 40 female clerical workers [Evans and Johnson 2000]. The women who were working during exposure to simulated open-office noise (55 dBA, peak level of 65 dBA) made fewer adjustments to the work station (chairs, foot rests, whiteboards and document holders) than women who worked in quiet conditions. Noise can affect performance also in other ways by masking information, leading to distraction, alter the arousal leaval and change the strategies chosen to perform a task [Kjellberg and Landström 1994]. Strategies chosen to perform a task can be to focus on the most important part of a task at the expense of less important ones, meaning a selective intake of information. Another strategy can be to add extra effort so that the performance level continues unchanged. However, this extra effort may result in as reduced efficiency and as an increase in adverse effects after the termination of the noise exposure. Due to the sound pressure level and characteristics of the low frequency noise in this thesis, performance effects can be expected to be within the theoretical explanations of the arousal theory, the information overload theory and strategies chosen to perform a task. A noise effect does, as mentioned above, not necessarily occur only during the noise exposure. Frankenhauser and Lundberg [1974] found that noise (using noise bursts of “everyday sounds” at 65-85 dBA, aperiodic noise) during 80 minutes of work with an arithmetic task had adverse after-effects, even if no effects were seen on performance during the noise exposure. Two studies have found that noise exposure leads to after-effects in the form of decreased motivation [Evans and Johnson 2000; Boman and Hygge 2000]. In the study by Evans and Johnson [2000] it was found that personnel who had been working for three hours with clerical tasks during exposure to simulated open office noise (55 dBA, 65 dBA peak) made fewer attempts to solve insoluble puzzles as compared with personnel working in a control condition described as a no-noise condition (ambient level of 40 dBA). 22

Boman and Hygge [2000] found that motivation in an insoluble embedded figures task was lower after a 15-minutes exposure to unpredictable noise (a conglomerate of noises at a maximum level of 76 dBA) than after exposure to predictable noise. The results of these two studies are in accordance with the theories on after-effects of noise exposure described by Glass and Singer [1972]. They stated that noise effects do not necessarily occur during the noise exposure itself but more often after termination of the noise exposure. Those effects have rather consistently been found in tasks such as insoluble puzzles performed after the noise exposure, i.e. tasks that do not primarily measure the capacity to perform but rather the motivation to perform well. Exposure to unpredictable and uncontrollable noise can also make a person less helpful when the experiment is terminated [Cohen 1980]. In summary, previous studies suggest that low frequency noise may impair performance. A number of hypotheses on possible mechanisms are outlined, but how these mechanisms function during exposure to low frequency noise needs to be explored.

Subjective effects

Symptoms

In a field study among 909 persons who answered a health questionnaire, the most frequently reported symptoms due to exposure to low frequency noise and infrasound were irritation, headaches and ”head feels heavy” [Nagai et al. 1989]. Similarly, it has been found that persons who reported themselves to be annoyed by exposure to low frequency noise in their homes, from heat pumps and/or ventilation installations, also reported a higher occurrence of headaches, sleep disturbance and psychosocial symptoms [Persson Waye and Rylander 2001]. In a recent study among control room workers who in a questionnaire made complaints about noise at the workplace, problems with concentration, drowsiness and headaches were reported due to noise exposure [Pawlaczyk-Luszcynska et al. 2002]. The noises in the control rooms comprised low frequency components (range 47.7-60 dB LAeq and 59.4-79 dB LCeq). A number of studies have found that exposure to low frequency noise and/or infrasound can

23

result in reduced wakefulness measured in a reaction-time task sensitive to tiredness [Kjellberg et al. 1998] or recordings of, in most cases, EEG [Landström et al. 1982; Landström et al. 1983; Landström 1987; Landström 1990]. Subjectively reported reduced wakefulness or increased tiredness was described in a field study of 439 persons working in offices, laboratories and industries [Tesarz et al. 1997]. A higher degree of subjective tiredness was found after work and a higher degree of annoyance was reported by the employees in the noise environments where the difference between the C- and A-weighted sound pressure levels was more than 15 dB, which was used as an indicator of low frequency noise, than by employees at workplaces where the difference between the C- and A-weighted sound pressure levels was smaller. The number of subjects working in low frequency noise was rather small (n=15), however, and no corrections were made for other factors that could have affected tiredness after work. The symptoms of tiredness and motivation are closely related. It has been shown, for example, that sleepy persons are more likely than rested persons to use excuses for not working hard enough on a task [Kjellberg 1975] and that effects of fatigue on performance can be reduced by increasing motivation, e.g. by giving feedback on results or in other ways making a task more interesting [see review by Kjellberg 1977]. It is possible that tiredness and motivation can be negatively affected by exposure to noise, and this can then probably lead to effects on performance. It is thus also possible that tiredness may reduce motivation and that greater motivation may reduce tiredness. According to Evans [1998], motivational after-effects are more likely to occur if subjects experience the prior noise exposure as uncontrollable. Regarding low frequency noise, there is a lack of studies evaluating whether low frequency noise could affect motivation and/or induce after-effects. As it is possible that tiredness can be an effect of exposure to low frequency noise, it can be hypothesised that motivational effects also could appear.

24

Annoyance

At the present, noise annoyance is the most widely used criterion for adverse effects of noise, excluding high sound pressure levels, and can be described as a feeling of displeasure evoked by a noise. Annoyance is closely related to feelings described by the words: disturbance, dissatisfaction, concern, bother, irritation, nuisance, discomfort, uneasiness and distress [see e.g. Guski 1999; Guski et al. 1999]. Borsky [1972] defined annoyance in general to be “a feeling of displeasure associated with any agent or condition realized or believed by an individual or group to be adversely effecting them”. Explanatory factors for whether a person becomes annoyed when exposed to noise can be divided into the following categories: individual factors (permanent and temporary) such as hearing impairment, noise sensitivity, attitude to the noise source, physiological and psychological state; situational factors such as activities performed or intended to be performed; sound properties such as noise level and sound characteristics (i.e. fluctuation and tonality); and factors related to noise source such as controllability of the noise source, information content and permanence [see e.g. Jones and Davies 1984; Landström et al. 1999; Guski 1999; Guski et al. 1999; Hallmann et al. 2002]. Some studies have evaluated sound characteristics and their relation to annoyance and several of them have found that the A-weighted sound pressure level underestimates annoyance caused by low frequency noise [see e.g. Kjellberg et al. 1984, Kjellberg and Goldstein 1985, Persson et al. 1990; Persson Waye 1995; Persson Waye et al. 1997; Persson Waye and Rylander 2001]. On the other hand, Key and Payne [1981] compared a one-third-octave band at 1000 Hz with a low frequency noise, a high frequency noise and one-third-octave bands of pink noise (centred frequencies at 63, 125, 250, 500, 1000, 2000, 4000 and 8000 Hz), all at 90 dB SPL, and found that the high frequency noise was rated as most annoying. These noises in the latter study were presented for only two seconds, however, and it can thus be questioned whether annoyance or, which seems more likely, loudness was evaluated. Landström et al. [1997b] found that persons working in control rooms reported higher annoyance than persons working in offices. The reported annoyance increased somewhat during the week among the control room workers while it decreased somewhat among the office workers. The noise environment in the control rooms was dominated by low frequencies, while the noise in the offices was of a more varied frequency range, which could

25

explain why no relationship was found between the sound pressure level (measured in dBA) and annoyance when the 51 control rooms were compared with the 71 offices [Landström et al. 1997b]. Persson et al. [1990] compared noises with different kind of spectra and their relevance for annoyance. The results showed that when the 98 subjects were exposed to ventilation noises centred at 80, 250, 500 and 1000 Hz at the same A-weighted sound pressure level, the noise centred at 80 Hz was rated as more annoying at 60, 65 and 70 dBA. A field study by Pawlaczyk-Luszcynska et al. [2002] comprised a listening test with eight noises and was carried out among 215 men working in control rooms. The results showed that low frequency noises at comparable A-weighted sound pressure levels (range 48-66 dB) were rated as more annoying than broadband noises without a dominant content of low frequency components. Kjellberg et al. [1984] reported that, when 24 subjects adjusted the sound pressure level of two noises containing a high or low proportion of low frequencies in order to achieve the same level of annoyance, the A-weighted sound pressure level underestimated the disturbance of a low frequency noise by 5 dB at 50 dBA and by 8 dB at 86 dBA. This means that a noise containing low frequencies could have a noticeably lower A-weighted sound pressure level than a noise not containing low frequencies and still be equally annoying. Similarly, Byström et al. [1991] found that, when two groups of 24 subjects were instructed to tune in the “highest level at which it was possible to maintain the performance level without extra effort” of a noise with a middle frequency of 100 Hz or a noise with a middle frequency at 1000 Hz, the acceptable level for work performance was about 6 dB lower for the low frequency noise. In summary, low frequency noise may cause other subjective symptoms and higher ratings of annoyance than noises at comparable A-weighted sound pressure levels that are not dominated by low frequencies.

Stress induced by noise

The experience of stress may be compared with the experience of pain, meaning that it is difficult to measure and understand another person’s experience. Stress reactions can be described as emotional, cognitive, behavioural and physiological [Levi 2002]. An emotional

26

reaction manifests itself as anxiety, worry and depressed state of mind. Cognitive reactions are difficulties in remembering, learning or concentrating. Behavioural reactions can be taking medicines or drugs or drinking alcohol. Finally, typical physiological reactions caused by stress are rapid heart rate, faster breathing rate and greater muscle tension. Two commonly mentioned coping mechanisms for stress are “defence mechanisms”, where the person is active and tries to control the stressor by “fight-or-flight”, and the “defeat reaction”, where the person is passive and experiences lack of control, helplessness, inferiority, distress, anxiety and depression. Humans have two different psychoendocrine stress systems: the sympathetic adrenal medullary (SAM) system and the hypothalamic pituitary adrenocortical (HPA) system. Lundberg [1999] writes that the SAM system is normally activated by the active “defence mechanisms”, which results in a secretion of the catecholamines epinephrine (adrenalin) and norepinephrine (nor adrenalin). The HPA system is normally activated by the passive “defeat reaction”, which results in a secretion of glucocorticoids (e.g. cortisol). As epinephrine and norepinephrine only can be measured in blood or urine, the response cannot easily be followed at repeated intervals over time without stressful interruption of the subjects. Cortisol levels can be measured in blood, urine and saliva. Cortisol levels in saliva accurately reflect free, physiologically active cortisol in the circulation [Kirschbaum and Hellhammer 1994; Aardal and Holm 1995], and the levels show a rather fast response to stress, which makes cortisol a good marker for measuring stress in experimental sessions. Measuring cortisol in saliva is a preferable method, as saliva sampling is a simple and nonstressful procedure [Kirschbaum and Hellhammer 1994]; hence, a rise in cortisol concentrations due to the sampling procedure is avoided [Aardal and Holm 1995]. Cortisol, a glucocorticoid hormone, is the main secretory product of the HPA neuroendocrine system. Cortisol is related to important health mechanisms - it regulates carbohydrate and lipid metabolism, modulates the immune system and has an important influence on mood and cognitive processing [see e.g. Miller and Tyrell 1995; Plihal 1996; Kirschbaum et al. 1996]. Many of cortisol functions are associated with a marked circadian rhythm, regulated by the hypothalamic suprachiasmatic nucleus [Buijis et al. 1999; Born et al. 1999]. The lowest cortisol level is found during the first hours of the night; about 30 minutes after awakening the level increases markedly after which there is a continuous decrease over the remainder of the day, except for temporary increases after meal times [Kirschbaum and Hellhammer 2000].

27

Several studies have evaluated how cortisol levels can vary as a result of stressful events during work. A relationship was found between increased salivary cortisol concentrations and perceived stress among air traffic controllers during radar work sessions [Zeier et al. 1996]. Increased levels of salivary cortisol have also been found in football coaches as they watched their teams play [Kugler et al. 1996], and increased levels of cortisol in saliva were found in students undergoing assessment [Evans et al. 1994]. Other studies have evaluated cortisol levels among chronically stressed people. The secretion varies depending on whether the subject has been under stress for a long time (chronically stressed) or whether the subject is experiencing an acute reaction. Schultz and Merck [1997], cited in Schultz et al. [1998], found that chronically stressed subjects had a lower cortisol response in the evening after an exhaustive day at work than non-stressed individuals. Like Schultz and Merck [1997], Pruessner et al. [1999] found that teachers in a high burnout group had lower overall cortisol levels in the morning compared than those in a low burnout group. Some studies have evaluated effects of noise exposure on cortisol [for a review see e.g. Kirschbaum and Hellhammer 1994; Ising and Braun 2002]. Elevated cortisol levels in urine have been found after chronic noise exposure at levels of 85 to 95 dB LAeq [Melamed and Bruhis 1996]. When the persons were equipped with earmuffs, the cortisol levels were lower at the end of the work shift as compared to the normal work situation without earmuffs. Another study found increased cortisol secretion in saliva of subjects doing mental work during an exposure to white noise at 90 dBA as compared to subjects working in a quiet condition at 55-60 dBA [Miki et al. 1998]. Temporary elevations in saliva cortisol levels have also been found after exposure to intermittent pink noise at 75 or 80 dBA [Yamamura et al. 1982]. Furthermore, salivary cortisol and cognitive performance were compared between a four-hour mental activity session during exposure to environmental noise at 65-70 dB and a quiet control session with an optional activity [Bohnen et al. 1990]. Elevated cortisol levels were found after the mental activity session, and subjects with a high cortisol response performed less well on a divided attention task in that session. Other studies have found no effects on cortisol levels caused by noise exposure [e.g. Andrén et al. 1982; Ortiz et al. 1974]. Andrén et al. [1982] found no effect on levels of cortisol, adrenalin, nor adrenalin, prolactin or growth hormones after 20 minutes of exposure to industrial noise at 95 dBA when they studied 15 male subjects resting on a bed. Neither did Ortiz et al. [1974] find an effect on cortisol levels among 18 subjects exposed to aircraft turbine noise at 105-115 dBA. The authors in the latter study suggested that the lack of effect in their study was due to the time of 28

day that the measurements were made. The majority of studies evaluating effects of noise on cortisol have used rather high sound pressure levels, and it would therefore be of interest to study cortisol levels during exposure to moderate levels of noise, and especially during exposure to moderate sound pressure levels of low frequency noise. There is also a lack of studies evaluating cortisol in relation to annoyance. To my knowledge, three studies have so far evaluated effects of low frequency noise on cortisol [Osguthorpe and Mills 1982; Ising and Ising 2002; Persson Waye et al. 2003]. Osguthorpe and Mills [1982] found that exposure to continuous low frequency noise at 84 dBA for 24 hours or 90 dBA for eight hours altered the circadian pattern of cortisol in plasma. The subjects usually slept or read during the exposure, as they were restricted to sedentary pursuits. The cortisol levels were still elevated 24 hours after the onset of both noises, despite the fact that the subjects exposed to 90 dBA had been in quiet conditions for 16 h. As no effects were found on heart rate, blood pressure or catecholamine level, the results indicate that the stress response was mediated primarily by activity in the HPA system. Ising and Ising [2002] evaluated effects of road traffic noise with 24 h lorry traffic, generating a noise with a frequency spectrum with its maximum below 100 Hz (mean LFmax amounted to 53 dBA and 78 dBC). They found increased cortisol levels during the first half of the night among children exposed to road traffic noise as compared to children living in a quiet area. Persson Waye et al. [2003] exposed 12 students to low frequency noise (A-weighted sound pressure level 40 dB and C-weighted sound pressure level 68 dB) during sleep and used as a comparison quiet reference nights. The results indicated that low frequency noise during sleep attenuated the cortisol response after awakening. It was also found that lower cortisol levels in the morning were associated with lower activity and pleasantness. In summary, cortisol has been found to be related to psychological stress and negative affect. Previous studies on noise and cortisol response indicate that high noise levels can alter the cortisol levels, although these studies are not conclusive. Little is known about the effect of lower noise levels, and only a few studies have evaluated effects of exposure to low frequency noise.

29

Sensitivity to noise

Effects of noise on performance seem to be inconsistent when the results are studied on individual level. Several studies have evaluated the importance of different individual factors for the effects induced by noise. A selection of individual factors found to be of relevance are for example critical tendencies [Weinstein 1980], neuroticism [Belojević and Jakovljevic 2001], negative affectivity [Smith et al. 2000; Smith and Rich 2002] and noise sensitivity [see e.g. Stansfeld 1992; Belojević and Jakovljevic 2001]. Noise sensitivity is one factor that has been considered to be important for whether or not a person exposed to noise will be negatively affected or not. Noise sensitivity can be measured by self-reporting in primarily three different ways [Job 1999]. The first variant is one single question on noise sensitivity. The other two variants form a scale on noise sensitivity using either multiple questions about noise sensitivity or questions in which the subjects rate their reactions to various noise situations other than the target noise situation. The poorest performance results during exposure to noise have been found among subjects sensitive to noise [Belojević et al. 1992; Jelinkova 1988]. In Belojević et al. [1992], subjects who were highly sensitive to noise in general, measured by Weinstein’s noise sensitivity scale [Weinstein 1978], had the lowest performance accuracy and rated the highest annoyance under conditions of recorded traffic noise. Similarly, Jelinkova [1988] found that noise sensitive persons showed reduced working ability and attention when exposed to recorded traffic noise (at 75 dB LAeq) as compared to persons tolerant to noise. However, Zimmer and Ellermeier [1999] found only weak relationships between subjective noise sensitivity and performance of a serial-recall task during exposure to pink noise, to Japanese speech at 76 dBA or in quiet conditions. Noise sensitivity has also been found to be an important factor for e.g. annoyance [see e.g. Öhrström et al. 1988; Dornic 1990; Persson et al. 1990]. Weinstein [1978] found that students sensitive to noise performed less well and also became more disturbed by dormitory noise during the year than non-sensitive students, who showed no difference between the two measurements. Furthermore, the sensitive students rated themselves as having less intellectual ability, lower social competence and a strong desire for privacy. In another study, noise sensitivity was found to be of importance for noise annoyance

30

and was also found to be related to current psychiatric disorder and individual factors such as neuroticism [Stansfeld 1992]. It has previously been suggested that subjects sensitive to low frequency noise are not necessarily sensitive to noise in general, as measured by general noise sensitivity scales [Persson Waye 1995]. In one study, 12 persons who were low frequency noise sufferers were measured as having an increased respiration rate when they were exposed to low frequency noise or rattling noise in comparison with nine control subjects who did not suffer from exposure to low frequency noise [Yamada et al. 1986]. As only a small number of studies have evaluated sensitivity to low frequency noise, especially with respect to effects on performance and annoyance, it seems important to further investigate whether subjects categorised as sensitive to low frequency noise are especially sensitive to exposure to low frequency noise. In summary, noise sensitivity seems to be an important factor for whether noise exposure will give negative effects, and the question is raised as to whether there may be subjects who are particularly sensitive to low frequency noise.

Summary of this review

This review indicates that low frequency noise may induce adverse subjective effects that are different from those induced by noises that are not dominated by low frequencies. Whether, and in that case how, performance and annoyance may be negatively affected by exposure to low frequency noise during work in occupational environments is not well explored. Further systematic studies in real-life conditions or in laboratory situations that evaluate effects of moderate levels of low frequency noise during work are necessary. Furthermore, the review indicates that an A-weighted sound pressure level is a less suitable descriptor for predicting annoyance and other effects of low frequency noise.

31

AIMS

The aim of this thesis was to answer the following questions:

×

Can low frequency noise at moderate levels that occur normally in office areas and control rooms affect performance, and are the effects influenced by workload?

×

Does exposure to moderate levels of low frequency noise during work lead to effects on annoyance and other subjective reports?

×

Can exposure to moderate levels of low frequency noise during work result in increased stress?

×

Are subjects sensitive to low frequency noise and/or sensitive to noise in general a potential riskgroup for adverse effects of exposure to low frequency noise during work?

×

What is the influence of the sound characteristics frequency balance and modulation frequency on subjects´ perception of a pleasant low frequency noise?

32

METHODS

Scientific approach

All studies were laboratory experiments. The set-up of the test chamber as an office area or a control room, the noise exposures, the performance tasks and the instructions to the subjects were all designed to resemble normal working conditions in environments such as office areas or control rooms. Annoyance and subjective well-being were studied as subjective estimations using questionnaires, while effects on performance and cortisol concentrations were studied objectively using different kinds of performance tasks and measuring cortisol in saliva. The local ethics committee approved all studies.

Study designs

The designs of the experiments are summarised below.

Papers I and II The experiment reported in paper I had a 2 (noises) ×2 (phases) ×2 (sensitivity groups) factorial design with repeated measures in the first two factors and measures of independent groups in the sensitivity factor. In the analyses of two tasks, a fourth factor, time periods within the task, was added. The experiment reported in paper II had a 2 (noises) ×6 (cortisol samples) ×2 (sensitivity groups) factorial design with repeated measures in the first two factors and measures of independent groups in the sensitivity factor. A within-subject design was chosen to make it possible to compare the same subject in both noise conditions. Each subject was exposed to the low frequency noise and the reference noise, starting alternatively with the low frequency noise or the reference noise, in a

33

randomised order in two separate test sessions on two separate days and always in the afternoon. The total exposure time was on average two hours and ten minutes. During the test sessions, the subject worked with four performance tasks twice, once in the first phase of the test session (phase A) and once in the second phase of the test session (phase B). The tasks were high, moderate or low demand tasks, and half were of a cognitive verbal character. Subjective effects were evaluated using questionnaires completed before, during and after the test sessions. To assess stress, six saliva samples were taken before, during and after the test sessions, and the amount of cortisol was determined. The subjects were categorised as high- or low-sensitive to noise in general and high- or low-sensitive to low frequency noise in particular using questionnaires. In the following text, when discussing these two papers in general, paper I and paper II will be referred to as papers I&II. Paper I is referred to in discussions of the effects on performance, annoyance and other subjective effects, while paper II is referred to in discussions of the effects on cortisol secretion and subjective stress.

Paper III The experiment reported in paper III had a 2 (noises) ×2 or 3 (times) factorial design with repeated measures in the second factor (tasks, questionnaires and saliva samples) and measures of independent groups in the noise factor. A between-subject design was chosen since the tasks selected to evaluate motivation could not be repeated and to avoid a probable negative effect on motivation from participating twice in a nearly 4-hour long test session. The subjects were exposed either to the low frequency noise or to the reference noise. The results in two tasks carried out in the learning session were used to create, in a randomised way, two noise exposure groups as equal as possible with regard to performance. All test sessions were carried out in the afternoon, and the total exposure time was on average three hours and 40 minutes, but could have lasted up to five hours depending on whether a voluntary task was completed. In the test sessions, the subject worked with five performance tasks. Most of the tasks were carried out twice, once in the first part of the test session and once in the second part of the 34

test session. The tasks were low or moderately demanding, half being of a verbal character. Subjective effects were evaluated using questionnaires that were completed before, during and after the test sessions. To assess stress, three saliva samples were taken - before, during and after the test sessions, and the amount of cortisol was determined. The subjects were categorised as sensitivity to low frequency noise using questionnaires administered before the test session.

Paper IV In the experiment reported in paper IV, three sound properties were varied and compared to the original low frequency noise in a between-subject design. The sound properties were evaluated four times in a randomised order, two at a time, starting alternatively at the minimum and the maximum value in the predefined ranges. A within-subject design was chosen to make it possible to compare all the resulting sounds from the adjustment with the original low frequency noise. The test session lasted an average of one hour. A short questionnaire was filled out after the test session. The subjects were categorised as sensitive to low frequency noise using questionnaires administered before the test session.

Exposure noises

In the experiments described in papers I&II and III, two ventilation noises were used: one of a predominantly low frequency character, “the low frequency noise”, and one of a predominantly flat frequency character, “the reference noise”. The experiment reported in paper IV used only the low frequency noise. A recording of a ventilation noise of a rather flat frequency characteristic (measured in thirdoctave bands) was used to create the exposure noises. The noise from the ventilation installation was recorded in front of the inlet grids in a large ventilation system. A B&K 4165 microphone was used, positioned close to the grid. A measurement amplifier (B&K Nexus 2690 preamplifier) with flat frequency response down to 20 Hz was used and the noise was recorded on a Sony TCD-D7 DAT recorder. The recording was done during the night to avoid other disturbing sounds from the surroundings. This recorded noise is called the reference 35

noise. To create a noise with dominant low frequencies, white noise was added by a random noise generator and filtered by a resonance filter with a centre frequency of 31.5 Hz. Furthermore a sinusoidal tone at 31.5 Hz was added and amplitude-modulated with a modulation frequency of 2 Hz and a modulation degree of 100%. This was done to give the low frequency noise a rumbling characteristic, which is naturally present in many noises dominated by low frequencies [Broner 1994]. The processing of the sound was done using a digital sound processor system (Aladdin interactive workbench, Nyvalla DSP, Stockholm, Sweden). The equivalent third-octave band sound pressure levels for the noises were measured at the position of the subjects’ head. In papers I&II, the A-weighted sound pressure level of both noises was 40 dB, and the corresponding C-weighted sound pressure levels were 50 dB for the reference noise and 69 dB for the low frequency noise. In papers III and IV, the A-weighted sound pressure level of both noises was 45 dB, and the corresponding C-weighted sound pressure levels were 53 dB for the reference noise and 72 dB for the low frequency noise. Figure 2 shows the equivalent third-octave band sound pressure levels, based on twominute measurements, for the low frequency noise and the reference noise at the A-weighted sound pressure level 45 dB.

Figure 2. The equivalent third-octave band sound pressure levels for the low frequency noise and the reference noise at the A-weighted sound pressure level 45 dB.

36

The A-weighted sound pressure levels were chosen so that they would be representative of levels commonly occurring in working environments such as office areas and control rooms.

Performance tasks

Figure 3 shows the experimental set-up, i.e. the combination of task demands and pressure put on the subjects, used in papers I&II and paper III.

TASKDEMANDS DEMANDS TASK HIGH,MODERATELY MODERATELYAND ANDLOW LOW HIGH, DEMANDTASKS: TASKS: DEMAND

LOWAND ANDMODERATELY MODERATELY LOW DEMANDINGTASKS: TASKS: DEMANDING

Verbalgrammatical grammaticalreasoning reasoningtask task Verbal Proofreading task Proofreading task Short-termmemory memorytask task&&bulb-task bulb-task Short-term Simple reaction-time task Simple reaction-time task

Bulb-tasksolely solely Bulb-task Simple reaction-time task Simple reaction-time task Searchand andmemory memorytask task(SAM1) (SAM1) Search Embedded figures task Embedded figures task Nonsensequestionnaire questionnaire Nonsense Proofreading task Proofreading task

PRESSUREPUT PUTON ONTHE THESUBJECTS SUBJECTS PRESSURE HIGHPRESSURE: PRESSURE: HIGH work fastand and work asasfast correctly as possible correctly as possible

LOWPRESSURE: PRESSURE: LOW work their work atattheir own pace own pace

NOISEEXPOSURE EXPOSURE NOISE

40dBA dBA 40 Reference noise Reference noise and and Lowfrequency frequencynoise noise Low

45dBA dBA 45 Reference noise Reference noise or or Lowfrequency frequencynoise noise Low

WORKENVIRONMENT ENVIRONMENT WORK ”Officearea” area” ”Office

”Controlroom” room” ”Control

Figure 3. Experimental set-up used in papers I&II and paper III. The performance tasks used in the experiments reported in papers I and III were selected according to the hypothesis that exposure to low frequency noise during work can lead to specific performance impairments, reinforced by workload, i.e. task demands and pressure put

37

on the subjects. It was important that the tasks corresponded to relevant assignments or reflected relevant demands on performing the assignments well in the two different working environments. The reason for this was to be able to set the results, if possible, in a real-life context. A working condition with a high workload that required a high performance level was created in the experiment reported in papers I&II by choosing tasks with mostly high and moderate demands on cognitive processing. High pressure was put on the subjects by the instruction to work rapidly and correctly during the test sessions. The working condition could be described as corresponding to an office area. In the experiment reported in paper III, a working condition with a low workload and less stressful working environment was created by choosing tasks of a primarily monotonous, routine character, and low pressure was put on the subjects by the instruction to work at their own pace. The working condition could be described as corresponding to a control room. To reinforce the workload, the working environment was described to the subjects, and the experiment room furnished so that the subjects could feel that they were working in an actual working environment and not in a laboratory. A short description of each performance task is given below. The verbal grammatical reasoning task [Baddeley 1968] was a cognitively demanding verbal task, measuring logical thinking. Mean response times and the number of correct and false answers were recorded. The proofreading task [Landström et al. 1997a] was a moderately demanding verbal task that required concentration. Number of lines read and number of typographical errors detected, number of contextual errors detected, number of correct marks (contextual and typographical), erroneous marks and total number of marks (correct and erroneous) in the text were recorded. The short-term memory task was a moderately demanding task measuring short-term memory. Mean response times and the percentage of correct and erroneous answers were recorded. The bulb-task [Persson Waye et al. 1997] was a low demand task of a monotonous, routine 38

character that measured attention and short-term memory. Mean response time and correct and erroneous answers were recorded. The bulb-task was used as a secondary task together with the short-term memory task in paper I. The set-up, in which the short-term memory task was performed at the same time as the bulb-task, was designed to require the subject’s full attention and concentration. In paper III, the bulb-task was used alone. The purpose in paper III was to study how the response time changed over time while working with a task for long time periods that suddenly and unpredictably required a rapid and correct decision. In order to create a lower workload in the experiment in paper III compared to the experiment in paper I, the time during which no bulbs were lighted was extended. The simple reaction-time task [Gamberale et al. 1989] was a low demand task of a monotonous and routine character in which reaction-time was measured. The task has previously been used to evaluate effects on fatigue [Kjellberg et al. 1998]. The low memory load version, SAM1, of the search and memory task [Smith and Miles 1987], was a low demand task that has previously been used to assess effects on sustained attention when performing the task in different combinations of noise, night work and meals [Smith and Miles 1987]. The embedded figures task was meant to measure motivation, expressed as time spent trying to solve insoluble tasks. It has previously been used to evaluate effects on motivational mechanisms after noise exposure, depending on task load and the predictability of the noise [Boman and Hygge 2000]. The nonsense questionnaire was prepared by the author for the purposes of paper III, to measure motivational mechanisms. The simple reaction-time task, the short-term memory task and the verbal grammatical reasoning task involved working with a computer. The bulb-task involved processing different coloured buttons at a control panel. The proofreading task, the embedded figures task, the search and memory task (SAM1) and the nonsense questionnaire involved working with pen and paper.

39

Questionnaires

The experiment reported in papers I and III used several different questionnaires to evaluate the subjects’ subjective reactions. A short description of each questionnaire is given below. Questionnaire evaluating symptoms: Comprised questions evaluating the degree of headache, tiredness, lack of concentration, irritation, dizziness, sleepiness, resignation, pressure over the eardrum or head, nausea, motivation, irritation in eyes or throat, a sensation of unpleasant taste and the subjects’ perception of being in a sociable mood. The majority of the symptoms were chosen in accordance with previous findings of subjective effects of exposure to low frequency noise [Persson Waye 1995], while some of the symptoms, such as irritation in eyes or throat, a sensation of unpleasant taste, were included to check the validity of the questionnaire. Questionnaire evaluating annoyance and subjective effects on performance: Comprised questions evaluating annoyance caused by the noise, how the subjects performance had been affected by noise, temperature or light during the test session and how interesting and difficult the tasks were perceived as being and how much effort the tasks demanded. In the experiment reported in paper I, a similar question on effort was posed directly after completing each task. Questionnaire evaluating mood [Sjöberg et al. 1979]: Comprised 71 adjectives that described different feelings, classified in six mood dimensions: social orientation, pleasantness, activation, extraversion, calmness and control. Questionnaire evaluating stress and energy [Kjellberg and Iwanowski 1989]: Comprised six adjectives describing stress and six adjectives describing energy. In the questionnaires, the subjects were given verbal alternatives. These ordinal scales resulted in ordered data, and equal intervals between the alternatives can thus not be assumed. Although not entirely in agreement with statistical theories, parts of the analysis in this thesis adopted parametric methods. This was done only when data did not diverge from a normal distribution.

40

Saliva sampling and cortisol determination

The saliva samples were taken using the salivette saliva sampling device (Sarstedt Ltd, Leicester, UK). The subject was asked to chew on a sterile dental swab for exactly three minutes. Samples were kept at -70°C until analysis. To allow for a proper baseline level, the subjects were instructed to take nothing by mouth other than water for at least 45 min prior to and during the test sessions [see Kirschbaum et al. 1997]. In the experiments reported in papers II and III, the subjects rested for 20 minutes in a separate room before the first saliva sample was taken. This time period was considered sufficiently as the subjects were also instructed to avoid rushing to the laboratory. After the rest and the first saliva sample, the noise exposure and test session started. In the experiment reported in paper II, six saliva samples were collected. The first sample served as a baseline value (mean base) collected after the subjects had rested for twenty minutes. Four samples were then taken during the first phase of the test session, imbedded between the performance tasks to represent the perceived stress of each task. The sixth sample was taken at the end of the test session, representing the second phase of the test session. Three saliva samples were collected in the experiment reported in paper III. The first served as a baseline value (mean base) collected after the subjects had rested for 20 minutes. The second sample was taken in the middle of the test session, which was taken to represent the first part of the test session, and the third and final sample was taken at the end of the test session, representing the second part of the test session. Saliva cortisol was determined by ELISA, specifically designed for the assay of cortisol in saliva (Salimetrics). The test has inter and intra assay % coefficient of variance <10% [Doyle et al. 1996]. The analysis of the saliva samples in paper II was made by Angela Clow and coworkers at the Department of Psychology at the University of Westminster in London. The analysis of the saliva samples in paper III was made at the Department of clinical Chemistry at Sahlgrenska Hospital, Göteborg University. The same method was used at the two different sites.

41

Subjects

The subjects were recruited by advertisement at Göteborg University. They participated voluntarily and received financial compensation for their time. Each potential subject underwent a hearing test (SA 201 II Audiometer, Entomed, Malmö, Sweden) and only persons with normal hearing, <20 dB HL, were included in experiments. Nineteen female and 13 male subjects participated in the experiment reported in papers I&II (average age 23.3 years, Sd=2.58), 38 female subjects participated in the experiment reported in paper III (average age 24.6 years, Sd=4.24), and 20 female and ten male subjects participated in the experiment reported in paper IV (average age 25.6 years, Sd=4.98). The reason for including only female subjects in the experiment reported in paper III was that a previous study found gender differences in the embedded figures task [Boman and Hygge 2000]. Except for the experiment reported in paper IV, it was important that only non-smoking and non-snuff-using subjects were included.

Subjective sensitivity to noise

Persson Waye et al. [1997] found indications in a pilot study of impaired performance during exposure to low frequency noise in subjects categorised as high-sensitive to low frequency noise compared to when the same subjects worked during exposure to a reference noise. As the experiment reported in papers I&II were an extension of the work performed in the pilot study [Persson Waye et al. 1997] the same way of categorising the subjects as sensitive to low frequency noise was used. In the text, subjects categorised as highly sensitive will be referred to as “high-sensitive” and subjects categorised as less sensitive as “low-sensitive”. The subjects were, at a separate occasion before the test session, exposed to a high level of a low frequency noise for 15 minutes. Subjects who after the exposure reported a pressure over their eardrums were categorised as high-sensitive to low frequency noise. Among the subjects categorised in this way, an equal number of high-sensitive and low-sensitive subjects was included in the experiment reported in papers I&II. However, when the results of the performance tasks in papers I&II were analysed, the sensitivity categorisation was not related

42

to noise effects. Instead, there seemed to be a difference between noise conditions for subjects who rated themselves as high- or low-sensitive to low frequency noise and high- or lowsensitive to noise in general. The ratings were given in a questionnaire completed after the last test session, together with other questions regarding the test session and personal data. Further analyses were carried out in order to test whether the questionnaire could offer a better way to make the sensitivity categorisation. For the analysis of sensitivity to low frequency noise, two questions were used. The first question was “are you sensitive to low frequency noise”, with five response alternatives ranging from “not at all sensitive”, “not very sensitive”, “rather sensitive”, “very sensitive” to “extremely sensitive”. The second question was “I am sensitive to rumbling noise from ventilation systems”, with six response alternatives ranging from “disagree completely, “practically disagree”, “partly disagree”, “partly agree”, “practically agree” to “agree completely”, included at the end of the original Weinstein’s noise sensitivity scale [1978]. For the analysis of sensitivity to noise in general, the question “are you sensitive to noise in general”, with five response alternatives ranging from “not at all sensitive”, “not very sensitive”, “rather sensitive”, “very sensitive” to “extremely sensitive”, was used together with the total number of points scored on Weinstein’s noise sensitivity scale [1978]. The sensitivity categorisations were elicted through a principal component analysis with direct oblimin rotation of the four sensitivity questions. The two categorisations of sensitivity to noise in general and sensitivity to low frequency noise were virtually independent (chi2=0.508, p=0.473). The procedure for the analysis is described in detail in paper I. The categorisation into sensitivity to low frequency noise was chosen to be an individual factor important for effects caused by exposure to low frequency noise. It was decided to categorise all subjects in the subsequent experiments (papers III and IV) as sensitive to low frequency noise, corresponding to subjects high-sensitive to low frequency noise. To be consistent throughout the papers, all questions regarding sensitivity to noise, including Weinstein’s noise sensitivity scale [1978], were completed by the subjects. To make it possible to include only sensitive subjects in the experiments, the sensitivity questions were filled out before the test session. To mask that only some of the questions were important, the sensitivity questions were included in a questionnaire comprising six pages with different questions regarding the subject’s health and education. The questionnaire was distributed before the test session at which the subjects reported an interest in taking part in the 43

experiment. Consequently, only subjects categorised as sensitive to low frequency noise were included in the experiments reported in papers III and IV. The minimum and maximum possible scores on the two questions on sensitivity to low frequency noise were 2 and 11. The minimal value for inclusion was set to 6, as this was the breakpoint value used to categorise the selected group of subjects in papers I&II into a high- or low-sensitive group. Table 1 shows the mean values for the categorisation of sensitivity to low frequency noise in all papers. The average value in the first sensitivity question corresponded to “rather sensitive to low frequency noise” (average mean papers I&II 2.9 Sd=0.6; paper III 3.2 Sd=0.6; paper IV 3.1 Sd=0.6) and the second question corresponded to between “partly agree” and “practically agree” (average mean papers I&II 4.3 Sd=0.7; paper III 4.5 Sd=0.9; paper IV 4.2 Sd=0.9).

Table 1. Mean values for the categorisation into sensitivity to low frequency noise. Subjects categorised as (high)

Subjects categorised as low-

sensitive to low frequency noise

sensitive to low frequency noise

Papers I&II (n=32)

7.3, Sd=1.2 (range 6-10)

4.1, Sd=0.8 (range 3-5)

Paper III (n=38)

7.6, Sd=1.3 (range 6-11)

Paper IV (n=30)

7.4, Sd=1.3 (range 6 –10)

As can be seen in Table 1, no major difference between papers in average values for subjects categorised as high-sensitive/sensitive to low frequency noise could be detected. Thus the subjects had an equal sensitivity to low frequency noise in all the experiments.

Test chamber

The experiments were carried out in a 24 m2 room. The temperature was kept constant at 21 degrees Celsius. The A-weighted sound pressure level from the background noise due to the test chamber ventilation was less than 22 dB, and the sound pressure levels for frequencies

44

below 160 Hz were below the threshold of normal hearing [ISO 389-7, 1996]. The walls were partly covered with absorbing material. Four loudspeakers and one subwoofer, all placed hidden behind curtains, were used to reproduce the noise. In the experiment reported in papers I&II, the room’s furnishings were set to replicate an office area. In the experiments reported in papers III and IV, the room was set up to replicate a control room.

Learning session

A learning session was held on a separate day before all the main test sessions. In the experiments reported in papers I&II and III this occasion was used to help the subjects learn the procedures and practise on short versions of the performance tasks. Written and verbal instructions were given before each task. To minimise subjective influence caused by the attitude to the noise, by motivation and by the individual’s own level of expectations before the test sessions, the information about the experiment did not explicitly refer to noise exposure. In the experiment reported in paper IV, the learning session was used so that the subjects could practise how to make the noise adjustments. In contrast to the other experiments, the information about noise in the working environment and the aim of the study to create more pleasant working environments as regards noise exposure was emphasised. In the experiment reported in paper III, which had a between-subject design, the learning session was also used to create two noise exposure groups as equal as possible with regard to performance, based on the results of short versions of two performance tasks carried out in the session.

Statistical treatment

Different statistical tests were used to determine the influence of noise exposure, time and subjective sensitivity on the different performance tasks, subjective reports and cortisol values. The statistical analysis chosen for the experiments reported in papers I, II and III was primarily analyses of variance, ANOVA. ANOVA was used to evaluate the influence of noise exposure and to identify possible two-way interactions between noise and time and three-way 45

interactions between noise, time and subjective sensitivity. Student’s t-test for dependent data (papers I and II) and for independent data (paper III) was used when ANOVA was not suitable. To correct for baseline values on the subjective reports in the experiment reported in paper III, the questions were related and analysed in relation to the response given at the beginning of the test session. Pearson’s correlation analyses were used to identify relationships between performance results, subjective reports and subjective sensitivity. The cortisol levels in the experiments reported in papers II and III were expressed as square roots of raw data to counteract initially skewed distributions. The analysis of cortisol levels and subjective stress during the test sessions was made in relation to the initial sample, expressing the change as percentage above or below the population mean. Wilcoxon’s two-tailed signed rank test was used in the experiment reported in paper IV. A mean value based on the four adjustments done on each sound property was calculated for every subject. The groups’ median value for each sound property was then compared to the original low frequency noise. This was not possible for the adjustments of modulation frequency, as either a low or high value was preferred. These data are therefore mainly descriptive. All analyses were two-tailed, and a p-value below 0.05 was considered statistically significant. A p-value up to 0.10 is reported as a tendency. The p-values in the analyses of variance are based on degrees of freedom corrected with Greenhouse-Geisser epsilon, when appropriate. To control for multiple comparisons in the correlation analyses, a p-value below 0.01 was set as the limit for statistical significance. More detailed information on statistical treatment can be found in each paper. The statistical analyses were carried out using SPSS software (SPSS base 10.0 for Windows; SPSS base 11.0 for Windows).

46

MAIN RESULTS

Paper I

Performance

During work with the verbal grammatical reasoning task, no difference in total response time was seen between noise conditions in the first phase of the test session. The response time in the second phase was shorter in both noise exposures, as compared to the first phase (F(1,31)=9.014, p<0.01). The analysis further showed that there was an interaction between noise and phase (F(1,31)=5.750, p<0.05). The interaction showed that the decrease in response time in the second phase compared to the first phase was less pronounced during the low frequency noise as compared to the decrease during exposure to the reference noise. A tendency to an interaction between noise, phase and low frequency noise sensitivity was found (F(1,30)=3.319, p=0.078). Figure 4 shows that subjects high-sensitive to low frequency noise had a similar response time in both noise exposures in the first phase. In the second phase, the response time during exposure to the low frequency noise was nearly equal to the response time in the first phase. In contrast, the response time in the second phase during exposure to the reference noise was somewhat shorter than the response time in the first phase.

47

Figure 4. Average mean and Sd for response times in the verbal grammatical reasoning task in subjects categorised as high-sensitive (HS) and low-sensitive (LS) to low frequency noise (paper I). In the proofreading task, the analysis showed that there was an interaction between noise and phase for erroneous and total (including both correct and erroneous) marks (F(1,31)=10.069, p<0.005; F(1,31)=7.018, p<0.05). Fewer erroneous marks and fewer total marks were made in the second phase during exposure to the low frequency noise, compared to during exposure to the reference noise. Further analysis showed that there was an interaction between noise, phase and low frequency noise sensitivity in the number of lines read (F(1,30)=5.306, p<0.05). Subjects high-sensitive to low frequency noise read fewer lines over time in low frequency noise and a greater number of lines over time in the reference noise, while the reverse was seen for low-sensitive subjects. A similar interaction showing a somewhat different pattern was found also for subjects high-sensitive to noise in general (F(1,30)=7.976, p<0.01). No significant difference between the low frequency noise condition as compared to the reference noise condition was found in the simple reaction-time task or the bulb-task.

48

Subjective reports

The low frequency noise was rated as more annoying and was reported to impair the work capacity more than the reference noise (F(1,31)=9.922, p<0.005; F(1,31)=6.808, p<0.05). Subjects categorised as high-sensitive to low frequency noise gave a higher annoyance rating during work in the low frequency noise than the reference noise, while low-sensitive subjects rated both noises as equally annoying (F(1,30)=6.534, p<0.05). No main effect of noise condition was found on the subjective symptoms or the mood dimensions. Subjects highsensitive to low frequency noise reported a lower perception of “being in control” after than before exposure to the low frequency noise, while the opposite was found for low-sensitive subjects (F(1,29)=4.352, p<0.05). A tendency toward the same interaction was present for "activation", showing a lower value for activation during both noises for subjects highsensitive and subjects low-sensitive to low frequency noise, but the decrease was greater for high-sensitive subjects during exposure to low frequency noise (F(1,29)=3.837, p=0.06). Annoyance caused by low frequency noise was correlated to a feeling of pressure on the head, tiredness, dizziness and lack of concentration (rxy=0.664, p<0.001; rxy=0.519, p<0.005; rxy=0.519, p<0.005; rxy=0.537, p<0.005). Reference noise annoyance was correlated only to nausea (rxy=0.522, p<0.005). Reported impaired work capacity due to low frequency noise was correlated to lack of concentration, nausea, tiredness and a feeling of pressure on the head (rxy=0.507, p<0.005); rxy=0.460, p<0.01; rxy=0.471, p<0.01; rxy=0.494, p<0.005). No correlations between reported impaired work capacity and symptoms were found for reference noise. No significant relationships between symptoms and performance were found that were expected to be related to noise exposure. Some of the questions were included in the questionnaires to check validity, for example irritation in eyes or throat and a sensation of unpleasant taste, in the questionnaire evaluating symptoms. The results in these questions showed no difference between noise conditions.

49

Paper II

Stress indicator

During exposure to both noise conditions, the level of cortisol had decreased at the last sampling point compared to the mean base sample (F(1,29)=26.598, p<0.001). This was in accordance with expectations, based on the known circadian rhythm. Cortisol levels during the performance tasks were not significantly affected by noises or related to noise sensitivity alone. However, the normal circadian decline in cortisol level was less pronounced for subjects high-sensitive to noise in general when they were exposed to low frequency noise (F(1,28)=4.681, p<0.05). For subjects categorised as high-sensitive to low frequency noise, a tendency toward the same interaction, between noise, time and sensitivity, and a similar response pattern for cortisol levels were found (F(1,28)=3.736, p=0.063). Figure 5 shows the average values of cortisol during exposure to the low frequency noise for subjects high- and low-sensitive to noise in general. Figure 6 shows the average values of cortisol during exposure to the reference noise for subjects high- and low-sensitive to noise in general.

50

Figure 5. Averge mean and SE for cortisol levels during exposure to the low frequency noise, subjects high-sensitive (HS) and low-sensitive (LS) to noise in general (paper II).

Figure 6. Averge mean and SE for cortisol levels during exposure to the reference noise, subjects high-sensitive (HS) and low-sensitive (LS) to noise in general (paper II).

51

As can be seen in Figure 5, subjects high-sensitive to noise in general maintained higher cortisol levels, relative to mean base values, during exposure to the low frequency noise as compared to low-sensitive subjects. Figure 6 shows an opposite but less pronounced pattern during exposure to the reference noise. Subjective stress was highest during both noise conditions at the times of the second and fourth questionnaires (F(1,28)=15.066, p<0.001), in accordance with reported effort in the different performance tasks. In general, subjects reported higher subjective stress during reference noise than the low frequency noise (F(1,28)=11.313, p<0.005). There were no clear correlations between cortisol levels and subjective stress or performance. The perception of a lower value of “being in control” after exposure to low frequency noise was correlated to a higher value of subjective stress for the whole group (rxy=-0.582, p<0.001).

Paper III

Performance

The results of the bulb-task are shown in Table 2. A lower number of bulbs were responded to correctly in the low frequency noise condition than when working with the task during exposure to the reference noise (F(1,35)=4.345, p<0.05). The analysis showed a significant interaction between noise and time on erroneous response time (F(1,36)=6.586, p<0.05). Subjects in low frequency noise increased their erroneous response time over time while subjects in reference noise decreased their erroneous response time over time. The result on the correct response time showed a tendency towards the same pattern (F(1,34)=3.205, p=0.082).

52

Table 2. Results in the bulb-task (paper III). Reference

Low frequency

noise

noise

p-value

Second Main effect

p-value

First

Second

First

Interaction

time

time

time

time

(noise)

(noise*time)

Erroneous response time (s)

2.1

1.8

2.1

2.4

p=0.111

p<0.05

Correct response time (s)

1.9

1.8

2.0

2.1

p=0.172

p=0.082

Number of correct responses

14.7

14.8

13.8

13.6

p<0.05

p=0.591

The results in the search and memory task (SAM1) are shown in Figure 7. There was an interaction between noise and time in the number of lines searched (F(1,36)=7.828, p<0.01). The number of lines searched the first time the task was carried out was about equal for both noises. The subjects increased their working speed from the first to the second time during both noise conditions; however, the increase in working speed was lower when working with tasks during exposure to the low frequency noise than working during exposure to the reference noise.

Figure 7. Average mean and Sd for number of lines searched in the search and memory task, SAM1 (paper III).

53

In the proofreading task, there was a tendency to significance in subjects reading a larger number of lines and making fewer erroneous marks per line during exposure to the low frequency noise compared to the reference noise (F(1,35)=3.774, p=0.060; F(1,35)=2.947, p=0.095). There was an interaction between noise and time on number of typographical marks per line (F(1,35)=4.654, p<0.05). Subjects working in low frequency noise made on average an equal number of typographical marks per line the first and the second time the task was carried out compared to the subjects who worked in the reference noise condition. The subjects working in reference noise found a greater number of typographical marks per line the second time the task was performed. No significant differences between the low frequency noise condition as compared to the reference noise condition were found in the simple reaction-time task, the embedded figures task or the nonsense questionnaire.

Subjective reports

The subjects in the low frequency noise condition rated the bulb-task as less difficult and less effort demanding than the subjects working with the task in the reference noise condition (t=2.041, p<0.05; t=2.072, p<0.05). Subjects working in low frequency noise reported higher values of irritation than subjects working in the reference noise (F(1,35)=5.802, p<0.05). No differences between noise conditions were found on annoyance, reported impaired work capacity due to noise or motivation. However, noise annoyance in low frequency noise was found to be related to lack of concentration, sleepiness and pressure in the eardrums (rxy=0.616, p<0.005; rxy=0.672, p<0.005; rxy=0.621, p<0.005). No relationships of this kind were found for reference noise.

Stress indicators The level of cortisol was similar during both noise conditions. The cortisol levels decreased in comparison with the mean base sample, which was in accordance with expectations, based on the known circadian rhythm (F(1,36)=20.420, p<0.001). There was no difference between

54

noise conditions on subjective stress and no relationships between cortisol levels and reported stress or performance.

Paper IV

Sound characteristics

The results of the adjustments of the modulation frequency showed that all four adjustments differed from the modulation frequency of 2 Hz in the original low frequency noise (Z=3.663, p<0.001; Z=-3.669, p<0.001; Z=-3.779, p<0.001; Z=-2.710, p<0.01). Two different groups appeared, one preferring a higher modulation frequency (mean 9.1 Hz) and one preferring a lower modulation frequency (mean 0.4 Hz,) compared to the original low frequency noise. The middle range was avoided by most subjects. The results of the adjustments II and III (without modulation) of the frequency balance are shown in Figure 8. In these adjustments, an increase of the levels of frequencies above 630 Hz would reduce the levels of frequencies below 500 Hz. The resulting sound comprised a lower content of low frequencies than the original low frequency noise (Z=-4.135 p<0.001). When the same adjustment was carried out without modulation in the original low frequency noise, a tendency towards the same pattern was seen (Z=-1.903, p=0.057).

55

Figure 8. Resulting sound from the adjustments of the frequency balance (paper IV).

In adjustments IV and V (without modulation), an increased level in the frequency range of 50 to 6300 Hz did not affect the dominant low frequencies at 31.5 Hz and below, but the Aweighted sound pressure level could increase from 45 dB to 65 dB. A minimum change of 2 dB, i.e. above 47 dB, was concidered as necessary to be perceptual. For both adjustment IV and V, the resulting sounds had an A-weighted sound pressure level below 47 dB (Z=-2.110, p< 0.05; Z=-3.466, p<0.001). The median A-weighted sound pressure levels were for adjustment IV 45.3 dB (25th perc – 75th perc; 45.1-47.2) and for adjustment V 45.1 dB (25th perc – 75th perc; 45.0-45.4).

56

DISCUSSION

Background

This thesis endeavoured to evaluate the effects of moderate sound pressure levels of low frequency noise during work in environments such as office areas and control rooms. Noise sources generating low frequency noise in these kinds of working environments are e.g. ventilation, heating and air-conditioning systems, computer network installations and compressors. Relevant variables for study were performance, subjective effects (e.g. annoyance and symptoms), noise induced stress (saliva cortisol and subjective stress), subjective sensitivity to noise and sound characteristics in the noise of relevance for pleasantness.

Methods

The effects studied in this thesis were acute effects caused by working with performance tasks during exposure to noise. All experiments were performed in laboratory conditions, and the relevance of the results for working conditions in real life must therefore be evaluated with care. It is important to bear in mind the influence of the experimental design as the alterations in performance found in experimental conditions could include biases, e.g. induced by the experimental situation [Rylander and Persson Waye 1997]. There are however advantages in laboratory experiments compared to field studies. The major advantage is that the researcher can control the various parameters of interest, such as noise exposure and kind of task. In this thesis, great effort was made to design the laboratory experiments so that they would be as comparable as possible to a real life working environment. The noise exposure, sound pressure level, performance tasks, the instructions to the subjects and the furnishing of the experimental room were all chosen to create conditions that could be representative for a normal workday in an occupational environment like an office area or a control room.

57

The general aim of the thesis was to study the influence of low frequency noise during work. Thus, the focus was on the effects induced by the combination of working with different levels of cognitive demanding performance tasks during exposure to low frequency noise. One reflection may be whether the subjective effects, such as annoyance and subjective symptoms, could have occurred during the noise exposure even without working with the performance tasks. This issue was of minor relevance in this thesis, however, as the aim was to evaluate effects of low frequency noise during work. All experiments used the same original low frequency noise, while the A-weighted sound pressure level and the content of modulations differed between experiments. The low frequency noise was compared to a reference noise with a rather flat frequency spectrum at the same A-weighted sound pressure level in order to compare a noise dominated by low frequencies to a noise not dominated by low frequencies. It is rather common to compare the specific noise exposure in experiments with a “quiet” condition without noise exposure, but, as most work environments today have some kind of noise exposure, this would have induced an unnatural situation that in itself could have induced a bias. The sound pressure levels in the experiments were rather moderate levels compared to many other studies. When comparing different studies, it is important to note the sound pressure level that has been used [Smith and Jones 1992]. In this thesis, the A-weighted sound pressure levels were 40 dB (papers I&II) and 45 dB (papers III and IV), sound pressure levels that other studies have referred to as “the quiet control condition”. The sound pressure levels were chosen to obtain ecological validity, based on the A-weighted sound pressure levels recommended for work environments like offices areas and control rooms. The highest recommended A-weighted sound pressure level for working in environments such as control rooms are 60 dB [AFS 1992:10]. 60 dBA was however considered by the author to be a very high level, and the A-weighted sound pressure level 45 dB was thus chosen as a more comfortable sound pressure level in which to work. The exposure time was chosen to be sufficiently long so that the results could, if possible, be applicable for a whole work day. In the experiment reported in papers I&II, the subjects worked for about two hours and, in the experiment reported in paper III, the subjects worked for nearly four hours in the afternoon, which can be approximately representative for the second half of a normal working day. Employees in an office area or a control room work with several kinds of tasks during a work day. Thus, great effort was made to choose relevant tasks that reflect real assignments or 58

assignments reflecting relevant demands in these work situations. To make it more realistic, it was also important to put pressure on the subjects to emphasise the workload. This was important in order to put the results, if possible, in a context of real work life. The subjects included in the experiments were young, healthy and motivated students at Göteborg University. However, the homogeneity of the group could be a limitation in the results. On the one hand, the effects may have been worse if the sample had been based on subjects from the normal work population. On the other hand, it has been found that noise is particularly disturbing when learning a new task and hence the effects may have been smaller if experienced personnel had taken part. The results of the experiment reported in paper I indicated that sensitivity to low frequency noise was partly different from sensitivity to noise in general. Subjects high-sensitive to low frequency noise performed less well and reported higher annoyance than subjects categorised as high-sensitive to noise in general. It was therefore hypothesised that subjects high-sensitive to low frequency noise could be a risk group for exposure to low frequency noise during work, and subsequent experiments (papers III and IV) were limited to include only subjects categorised as sensitive to low frequency noise, corresponding to high-sensitive to low frequency noise in paper I. The focus on subjects sensitive to low frequency noise could be viewed as a limitation, however, as it was no longer possible to compare subjects who were sensitive to low frequency noise with less sensitive subjects. The categorisation of subjects as sensitive to low frequency noise was originally based on a factor analysis of a group of students in the experiment reported in papers I&II, where subjects high-sensitive to low frequency noise represented about 56% of the selected group. This proportion is comparable with 53.5% among all subjects (n=458) who during three years (2000–2002) reported an interest in taking part in the subsequent experiments. It can thus be concluded that, in this particular group, half of the subjects were categorised as being sensitive to low frequency noise, meaning that the group of subjects studied did not represent a minority. However, it is not known how this estimation corresponds to the proportion in the whole population. Throughout the day, cortisol levels normally follow a gradual circadian declining pattern. In paper II, about half of the test sessions started at 12.30 and the other half at 15.00. The proportion of test sessions starting at the different times was similar for the low frequency noise and reference noise, and no statistically significant difference in cortisol level was found 59

between subjects starting the test session at 12.30 and those starting at 15.00. In the experiment reported in paper III, all test sessions started at 14.00. No control was made for females’ menstrual cycle or use of contraceptive hormones in the analysis of cortisol. In a recent study by Kudielka and Kirschbaum [2003], the authors pointed out that no precautions are needed to control for the menstrual cycle phase when studying cortisol levels in the awakening response. However, women taking oral contraceptives have been reported to have a tendency to a smaller cortisol decrease in the awakening response compared to women who do not use these contraceptives [Pruessner et al. 1997]. The effect of oral contraceptives was similar to the effect of gender. In the experiment reported in paper II in this thesis, no gender difference was found in cortisol levels. It was therefore judged that the influence of the menstrual cycle or use of contraceptive pills could be considered to be of minor relevance for cortisol levels during the day. Saliva cortisol response has a time lag of five to 20 minutes, with maximal cortisol levels ten to 30 minutes after the end of the stress episode [Kirschbaum and Hellhammer 2000]. If there is a prolonged period of stress, the cortisol secretion can go on for several hours. Kirschbaum and Hellhammer [2000] gave the example of marathon runners: the runners’ cortisol levels in saliva were observed to increase uninterrupted for over four hours and to peak approximately 30 min after the run was finished. The sampling intervals in the experiments in this thesis were chosen to reflect the stress elicted by the different performance tasks. In paper IV, the subjects could interactively vary different parameters in the noise to make it more pleasant. Three subjects were influenced by the starting values when adjusting the modulation frequency. This behaviour was not found in the adjustments of the frequency balance, probably because adjustments of spectral balance are not perceived to have a clear high and low value. Furthermore, some subjects chose both very high values and very low values when adjusting the modulation frequency. When the author listened to the resulting sounds after the adjustment of the modulation frequency, the actual perceivable difference between the higher modulation frequency and the lower modulation frequency was very slight. This behaviour could thus be interpreted as both choices representing the ranges in which the modulation was less clearly perceived, meaning that both a very low and a rather high modulation frequency seem to be acceptable.

60

Results

Effects of low frequency noise on performance

The results of the performance tasks showed that the effects found during exposure to low frequency noise should be seen in the context of workload, i.e. task demands and pressure put on the subject. Two findings concerning performance were of the greatest importance. First, low frequency noise was found to impair performance in tasks with high and moderate demands on cognitive processing when performed under high workload. Tasks that had low demands performed under high workload were not negatively affected. The second finding was that low frequency noise impaired performance on some of the low demand tasks and the moderately demanding verbal task when performed under low workload, while tasks evaluating motivation were not affected. The effects found on the high demand task could be interpreted as being in accordance with the information overload theory. The subjects’ performance on the verbal grammatical reasoning task, a task that in itself required the subjects’ full work capacity in order to be performed well, did not improve over time during exposure to the low frequency noise as in the reference noise. The explanation for this may be the extra load of performing the task during exposure to low frequency noise, indicating that, in that situation, there was not enough spare mental capacity to learn the task equally well as when working with the task during exposure to the reference noise. A tendency towards the same result was found by Persson Waye et al. [1997] in the pilot study preceding paper I, using the same task and the same exposure noises and A-weighted sound pressure levels as used in in the experiment reported in paper I. The results are also in agreement with the results of a lower learning rate in this task during exposure to a simulated ventilation broadband noise with dominant energy in the lower frequency bands (51 dBA) [Kjellberg and Wide 1988]. Similar results were found by Benton and Leventhall [1986] when they compared exposure to low level low frequency noise (pure tones centred at 40 Hz and 100 Hz), to a narrow band noise (centred at 70 Hz), all at a level of 25 dB above the individual hearing threshold, and recorded traffic noise (90 dB Lin) and a silent control condition. A larger number of errors was found on a dual task during exposure to low level low frequency noise, and the differences between noise 61

conditions were more pronounced over time. The effects found on the low demand monotonous routine tasks, the bulb-task and the search and memory task (SAM1), can be interpreted as being in accordance with the arousal theory. As described earlier, the optimal level for successfully completing a simple task is higher than the optimal level for successfully completing a more complex task [Jones and Davies 1984]. Monotonous, machine-paced tasks, such as signal monitoring tasks comparable with the bulbtask, are sensitive to wakefulness changes [Hockey 1986] and low frequency noise has been found to reduce wakefulness or increase tiredness [see e.g. Landström 1990; Kjellberg et al. 1998]. The results found here can be explained by decreased attention and/or increased tiredness, which manifested itself as poorer performance results during work with tasks of a monotonous routine character. The results support the hypothesis that exposure to low frequency noise can result in increased tiredness. However, increased tiredness during exposure to low frequency noise was not supported by the separate subjective reports. The effects of decreased attention and/or increased tiredness were seen in the bulb-task, where the subjects during exposure to the low frequency noise needed a longer response time to make decisions, whether correct or incorrect, and, despite this longer response time, gave a greater number of erroneous answers. In the search and memory task (SAM1), it has previously been found that subjects generally work faster over time through practice and learning [Smith and Miles 1987]. However, subjects working during exposure to low frequency noise were found to show poorer improvement over time, i.e. a smaller increase in the number of lines searched as compared to subjects in the reference noise. The moderately demanding proofreading task is designed to measure typographical and contextual errors and number of lines read. The typographical errors require a superficial processing of the task while the detection of the more complex contextual errors requires a deeper processing and an understanding of the text [Kjellberg and Landström 1994; Weinstein 1974]. For this reason, it is necessary to describe all findings in the task in order to identify whether the noise exposure made the subjects change their way of working with the task. Rather inconsistent results in the proofreading task have been found when the task was used to measure effects of noise exposure [see for example Jones et al. 1990; Weinstein 1974; Holmberg et al. 1993]. In this thesis, during exposure to low frequency noise in the high workload condition, the whole group made fewer erroneous and fewer total marks (including both correct and erroneous marks) per line read. In the low workload condition during 62

exposure to low frequency noise, the subjects generally read a somewhat greater number of lines, made fewer erroneous marks per line read and found fewer of the typographical errors than subjects who worked with the task in reference noise. No difference between the noise conditions was found in contextual errors in the low workload condition. No analysis was made of contextual and typographical marks in the high workload condition. Taken together, during exposure to low frequency noise, subjects read a larger number of lines and made fewer erroneous marks, but they failed to find those errors in the text they were instructed to find. The results thus seem to be related to a less thorough treatment of the text material when working with the task during exposure to low frequency noise. The strategy adopted seems to be similar under both high and low workload. Smith and Jones [1992] pointed out that the specific effects that noise can have on performance depend on type of noise exposure, kind of task performed, task demands and the subject’s individual characteristics. Tafalla and Evans [1997] stated that it is possible to maintain performance under stressful working conditions, at least for up to 90 minutes. However, the results in the high workload condition suggest that working with tasks of a high and moderately cognitive demanding character during high workload resulted in decreased performance within two hours when exposed to low frequency noise. The strength in this thesis was the combination of task demands and pressure put on the subjects in order to identify and clarify whether exposure to low frequency noise during work would result in negative effects on performance and how this is related to workload. The results showed that performing high and moderately cognitive demanding tasks under high pressure or performing low and moderately demanding tasks under low pressure during exposure to moderate levels of low frequency noise resulted in negative effects on performance. When summing up the adverse effects found on performance due to exposure to low frequency noise, the effects in absolute terms, expressed in e.g. seconds, were rather modest. However, the size of the effect does not automatically mean that the effects are without practical relevance. Missing one or two signals and/or having a longer response time can in certain situations be very critical.

63

Effects of low frequency noise on annoyance and other subjective reports

The results from the questionnaires showed that low frequency noise was rated as having a greater impariment on the work capacity and as being more annoying than the reference noise. The difference between noise conditions was significant in the high workload condition. The results are compared to the pilot study carried out by Persson Waye et al. [1997], here referred to as the pilot study, as this study was a preceding study to paper I. The subjects generally rated the low frequency noise as causing a greater impairment in work capacity than the reference noise. The difference between noise conditions was significant in the high workload conditions (paper I and the pilot study) but not in the low workload condition (paper III). Figure 9 shows the reported impairment of the work capacity. It can be seen that, despite the absence of a significant difference between noise exposures in the low workload condition, the values of the ratings are rather similar as compared with the ratings made in the high workload conditions. Furthermore, it can be seen that higher ratings were consistently given on low frequency noise compared to reference noise.

Figure 9. Average means and Sd of reported impaired work capacity due to low frequency noise and reference noise (paper I, paper III and the pilot study).

64

Under the assumption that exposure time and A-weighted sound pressure level would increase the rated impairment of the work capacity, one could have expected higher ratings in the low workload condition, as the exposure time was twice as long and the A-weighted sound pressure level was 5 dB higher. This was not found, however. Thus this indicates that there was no clear indication of the influence of workload imposed by exposure time and/or noise level. The most central factor in how the noise was reported to impair the work capacity was the actual noise. The subjects generally rated the low frequency noise as more annoying than the reference noise. The difference between noises was significant under high workload (paper I), but was not significant in the pilot study or in the low workload condition (paper III). Figure 10 shows the annoyance ratings.

Figure 10. Average means and Sd of annoyance due to the frequency noise and reference noise (paper I, paper III and the pilot study). A comparison of the results from the experiment reported in paper I with those of the pilot study, where the subjects were exposed to a similar A-weighted sound pressure level and worked for two hours and one hour respectively, implies that the noises were rated as more annoying when the exposure time was longer. The influence of exposure time could be in 65

accordance with Holmberg et al. [1993]. They found increased annoyance in relation to session length when subjects rated annoyance after five, 30 and 60 minutes of exposure to ventilation noises with different frequency spectra and at different noise levels (at 40, 35 and 20 dBA). In contrast, Poulsen [1991] found in a study evaluating the influence of time on annoyance caused by impulse noise or traffic noise (at 35, 50 and 65 dB LAeq) that session length (1, 5, 15 or 30 minutes) was of no importance for annoyance. The instructions in the latter study were to imagine the noise in the home environment; thus, the noise in the actual test environment was not studied. Despite the absence of a significant difference in rated annoyance between noises in the low workload condition, the overall comparison between the papers in the thesis shows agreement with Holmberg et al. [1993] and indicates the importance of exposure time for perceived annoyance. The difference between noises in rated annoyance was greater under high workload than low workload, and thus a significant difference between noise exposures in rated annoyance was only found when task demands and the pressure put on the subjects were high. Even if the subjects’ performance was negatively affected when working during exposure to low frequency noise in the low workload condition, the subjects seem not to have perceived the working situation as annoying. The results agree with those of Kjellberg and Landström [1994] and Landström et al. [1993]. Kjellberg and Landström [1994] suggested in their review that noise tends to be more annoying when complex tasks are being carried out. Landström et al. [1993] found when estimating annoyance thresholds for different kinds of tasks that the tolerance level was about 6 dB lower in a difficult reasoning task than in a simple reaction-time task. Furthermore, examining the experiment reported in paper I and the pilot study with the experiment reported in paper III, it is seen that the noises were genrally rated as more annoying when the A-weighted sound pressure level was higher. This is in accordance with Holmberg et al. [1993], who found that annoyance was lower when the Aweighted sound pressure level was decreased from 40 to 35 dB. Similar results were reported by Landström et al. [1991], who found that a ventilation noise at a 5 dB lower level was considered more acceptable. Regarding reported symptoms, only irritation was found to differ between noises, and only under low workload. Subjects working during exposure to low frequency noise reported a higher value in irritation during the test session than subjects in the reference noise. For the mood dimensions, subjects high-sensitive to low frequency noise reported under high 66

workload a lower perception of “being in control” after as compared to before exposure to the low frequency noise, while the opposite was found for low-sensitive subjects. There was also a tendency to a significantly greater decrease for the perception of “activation” for subjects high-sensitive to low frequency noise during exposure to the low frequency noise, compared to low-sensitive subjects. Tables 3 and 4 show the relationships between annoyance and subjective symptoms and the relationships between reported impairment of the work capacity and subjective symptoms. Annoyance caused by low frequency noise was correlated to several different subjective symptoms, while annoyance caused by the reference noise correlated only to nausea. The pattern was similar for high and low workload. Impairment of the work capacity was correlated to subjective symptoms in the high workload condition during exposure to low frequency noise. Table 3. Relationships between annoyance and subjective symptoms (papers I and III). Annoyance due to the

Annoyance due to the

low frequency noise

reference noise

Paper I

Paper III

Paper I

Paper III

p<0.005

p<0.005

-

-

Sleepiness

-

p<0.005

-

-

Pressure over the eardrums

-

p<0.005

-

-

Pressure over the head

p<0.001

-

-

-

Tiredness

p<0.005

-

-

-

Dizziness

p<0.005

-

-

-

-

-

p<0.005

-

Lack of concentration

Nausea

67

Table 4. Relationships between reported impairment of the work capacity and subjective symptoms (papers I and III). Reported impairment of

Reported impairment of

the work capacity due

work capacity due

to low frequency noise

to reference noise

Paper I

Paper III

Paper I

Paper III

p<0.005

-

-

-

Sleepiness

-

-

-

-

Pressure over the eardrums

-

-

-

-

Pressure over the head

p<0.005

-

-

-

Tiredness

p<0.01

-

-

-

Dizziness

-

-

-

-

p<0.01

-

-

-

Lack of concentration

Nausea

Taken together, it seems that the subjects interpreted reported impairment of the work capacity and annoyance as two different dimensions. The ratings of impaired work capacity seem to have reflected the subjects’ conception of their performance in the work condition. Annoyance, however, seems to be a broader concept which, besides the influence of workload, exposure time and sound pressure level, also reflected how the subjects felt. This is in accordance with the definition of annoyance described at the beginning of this thesis, showing that several factors besides the actual noise exposure affect a person’s perception of annoyance. Further studies are needed however before it is possible to answer the question of whether the subjects were annoyed because they perceived symptoms or whether the subjects perceived symptoms because they were annoyed.

Effects of low frequency noise on cortisol and subjective stress

The results of the saliva samples showed that working under high workload during exposure to low frequency noise resulted in elevated cortisol levels among subjects high-sensitive to

68

noise in general. The cortisol response seems to be part of a larger stress pattern involving subjective effects, although no clear relationship between cortisol levels and subjective stress was found. In the high workload condition (paper II), higher ratings in subjective stress were found when the subjects carried out the high demand tasks and these tasks were rated as requiring more effort than the other tasks. No effect on cortisol levels due solely to noise conditions was found. This is interpreted such that, when subjects worked under high workload during exposure to the reference noise, they seem to manage the work situation rather well. However, the analysis showed that in low frequency noise, the normal circadian decline in cortisol levels was less pronounced in subjects high-sensitive to noise in general, as compared to lowsensitive subjects. During work in low frequency noise, the subjects performed less well and gave ratings of higher annoyance and a lower perception of “being in control”. Thus, the work situation that was manageable during exposure to reference noise resulted in increased cortisol levels when the subjects were exposed to low frequency noise. This is supported by previous suggestions that an increased secretion of cortisol is associated with the “defeat reaction” [Lundberg 1999]. An elevated cortisol level could hence be the result of subjects during low frequency noise experiencing it as not possible to control the working situation. This is also in accordance with the phenomenon of “learned helplessness” described by Seligman [1975]. The difference in cortisol levels in the two noise situations was statistically significant, although the difference was relatively small. As mentioned earlier, it is well known that cortisol is related to important health parameters, and a shift in cortisol levels could therefore have a negative influence on health. It can be hypothesised that the influence could be larger if the elevation is repeated or continues for an extended time period. Thus, although the clinical significance of the results needs further investigation, cortisol could be seen as a relevant indicator for unsatisfactory working conditions. The pattern described was seen in subjects categorised as high-sensitive to noise in general. In subjects high-sensitive to low frequency noise, a tendency toward an interaction with a similar response pattern was found. However, one would expect the cortisol response to be even more pronounced in subjects high-sensitive to low frequency noise during exposure to low frequency noise, as these subjects performed less well and reported the highest annoyance 69

during exposure to low frequency noise. There is thus a question whether something besides the sensitivity categorisations was of greater importance for the secretion of cortisol during exposure to low frequency noise. To my knowledge, it has not previously been found that noise sensitivity is associated with increased salivary cortisol levels, although other personality traits have been found to be of importance. In one study of 120 subjects, negative affect was significantly associated with increased salivary cortisol levels induced by naturally occurring daily stressors [Smyth et al. 1998]. Thus, negative affect was related to higher cortisol levels while positive affect was related to lower cortisol levels. Furthermore, the cortisol response has been found to be related to the concept of one’s own competence, extroversion, social resonance and trustfulness, meaning that high-cortisol respondents found themselves to be less attractive than others, to have poorer self-esteem and to be more often in a depressed mood. They also reported more symptoms than low-cortisol respondents [Kirschbaum et al. 1995]. Accordingly, it is possible that the difference in cortisol response between the two noise sensitivity categorisations could be a result of other individual factors besides noise sensitivity, which unfortunately were not measured in the experiments in this thesis. Noise sensitivity was chosen as the factor that has most consistently been found to be of importance for reported noise annoyance [see e.g. Öhrström et al. 1988; Stansfeld 1992] and effects on performance [Belojević et al. 1992; Jelinkova 1988]. In the low workload condition (paper III), no significant difference in cortisol levels was found between noises. Figure 11 shows the cortisol levels for both noises in the low workload condition. The set-up for the experiment reported in paper III was designed to have a low workload and not to be stressful, and thus no difference between noises was expected. The finding during low workload thus agreed with the hypotheses. In comparison, Peter et al. [1990] found that psychological effort was needed to maintain enough alertness to correctly perform a monotonous task in a monotonous condition and that the effort resulted in monotonous stress, measured as biological load (ECG) and alertness (EEG). However, this study differed in many ways from the low workload condition. In the study by Peter et al. [1990], subjects were comfortably seated for 90 minutes with their eyes closed and, under extreme monotony, performed a paced secondary task (the train function safety circuit), where the aim was to test whether the task can guarantee that a train driver is capable of driving a train.

70

Figure 11. Averge mean and SE for cortisol levels during exposure to the low frequency noise and the reference noise (paper III). Lower ratings of subjective stress were found in the high workload condition during exposure to low frequency noise compared to reference noise. This was unexpected as performance was impaired to a higher degree when the subjects were exposed to low frequency noise. The higher ratings of subjective stress may be the result of a feeling of being able to work rapidly during exposure to the reference noise. In comparison, the feeling of not being able to work as rapidly in the low frequency noise may have been perceived more as not “being in control”. In the low workload condition, no effect of noise was found on subjective stress. No clear relationships between cortisol levels and subjective stress were found in this thesis; only one positive correlation between cortisol level and stress was found at one of the measured occasions in the high workload condition, and only for subjects high-sensitive to low frequency noise. Several studies have attempted to measure subjective stress using questionnaires and, as in this thesis, no definitive conclusion on subjective stress or a plausible relationship between subjective stress and different stress hormones can be drawn. No relationships between subjective stress and epinephrine or norepinephrine were found among subjects working with performance tasks during exposure to 85 dBA [Arvidsson 1975]. Furthermore, no relationship

71

between annoyance due to industrial noise at 85 dBA and stress hormones in plasma was found among 12 male students [Fruhstorpher et al. 1988], and no relationship between subjective stress and urinary epinephrine was found among 105 females exposed to 56 dBA in a simulated hospital environment [Topf 1992]. Although the participants in one study perceived work in the noise condition (55 dBA, 65 dBA peak) as noisier than those working under quiet conditions (ambient noise at 40 dBA), no effect on perceived subjective stress was found [Evans and Johnson 2000]. Furthermore, elevated urinary epinephrine levels were found, but no effects were seen on norepinephrine or cortisol levels. The authors explained the lack of findings in self-reported stress as habituation to the noise, since all the subjects were experienced workers, the exposure time was rather long (3 h) and a realistic office noise was used. Finally, Pruessner et al. [1999] found no effects on perceived stress on cortisol levels directly after awekening. Further studies using for example more refined stress questionnaires are needed to draw conclusions about what more precisely subjective stress is, how it can be measured and how subjective stress is related to stress hormones such as cortisol.

The influence of subjective sensitivity to noise for adverse effects

The subjects in the experiment reported in papers I&II were categorised as high- or lowsensitive to low frequency noise and high- or low-sensitive to noise in general. The two categorisations were found to be virtually independent, using a principal component analysis with direct oblimin rotation of the sensitivity questions. In the experiment reported in paper I, subjects high-sensitive to low frequency noise performed less well and reported the highest annoyance during exposure to low frequency noise, as compared to subjects categorised as low-sensitive to frequency noise and to subjects high-sensitive to noise in general. In the experiment reported in paper II, subjects highsensitive to noise in general had a less pronounced decline in cortisol levels, and a tendency towards the same pattern was found for subjects high-sensitive to low frequency noise. Thus, the categorisation of sensitivity to low frequency noise could be distinguished from the categorisation of sensitivity to noise in general. Furthermore, subjective sensitivity to low frequency noise was the individual factor that seemed to be of greatest importance for adverse effects of low frequency noise during work.

72

Sound characteristics

The experiment reported in paper IV was designed to test the two sound characteristics’ modulation frequency and frequency balance and their influence on the perception of a pleasant low frequency noise. The “original low frequency noise” was the same low frequency noise as in the experiments reported in papers I&II and III, however the Aweighted sound pressure level and the content of modulations differed between experiments. It was hypothesised that if subjects who are sensitive to low frequency noise adjusted the low frequency noise to make it more pleasant, the resulting sound could be perceived as less annoying. The design of the experiment reported in paper IV did not include evaluations of annoyance, however, so this question is left to be answered in future studies. The resulting pleasant sounds were adjusted to have either a higher or a lower modulation frequency compared to the modulation frequency in the original low frequency noise. The most likely explanation as to why the subjects avoided the middle range and instead chose either a high or low modulation frequency is that both these adjustments represent the ranges in which the modulation is less clearly perceived. This is in accordance with previous findings, where a modulation frequency of 4 Hz was found to be most perceivable [Zwicker and Fastl 1999] and with one study where a sound modulated at 2-3 Hz was rated as more annoying than both lower and higher frequencies [Landström et al. 1996]. Thus, the results indicate that clearly perceived modulation frequencies were considered less pleasant, and a pleasant sound in a working environment should therefore contain only little or no perceivable modulations. The resulting pleasant sound from the adjustment of the frequency balance, where the subjects adjusted the level of high and low frequencies, showed that the subjects preferred a lower relative content of frequencies below 500 Hz compared to the original low frequency noise. The effect was more modest when the low frequency noise did not comprise modulations, which was in agreement with the adjustments of the modulation frequency, showing that the occurrence of perceivable modulations had an important effect on the result. When the Aweighted sound pressure level could be altered when adjusting the frequency balance, the subjects preferred the initial A-weighted sound pressure level to a higher content of high frequencies.

73

The slope theory has been shown to be more complex than previously suggested. The theory was criticised by Goldstein and Kjellberg [1985], as they in contrast to Bryan [1976] found that annoyance increased with a decreasing slope. The major differences between these two diverse conclusions could be due to that Bryan’s [1976] findings were based on empirical data of noise disturbance from transportation, and interior noise in the work- and home environment, while Goldstein and Kjellberg [1985] carried out experimental studies using noises with broad-band spectra in the frequency range 15-8000 Hz. Furthermore, Goldstein and Kjellberg [1985] compared the different slopes (–3 dB, –6 dB and –9 dB) at different Bweighted sound pressure levels (66.4-81.2 dB). It is therefore questionable whether Goldstein and Kjellberg [1985] conclusion about the relationship between slope and annoyance could be drawn as the B-weighted sound pressure levels covaried with the steepness of the slope. In the experiment reported in this thesis, the adjustment where the A-weighted sound pressure level could be altered could be seen as a replication of the design in Goldstein and Kjellberg [1985]. Figure 12 is a schematic description of the slope obtained in this thesis.

Figure 12. A schematic description of the slope obtained in this thesis. The results indicated that the subjects preferred the low frequency noise with a steeper slope to a noise with a higher A-weighted sound pressure level, which thus is in agreement with 74

Goldstein and Kjellberg [1985]. The slope theory is based on the assumption that the slope of the frequency spectra is the same irrespective of A-weighted sound pressure level, e.g. dBA. In comparison, the slopes of the loudness curves are different depending on sound pressure level. The slope is less steep at higher loudness levels of the equal loudness level contours. Hellman and Broner [1999] compared absolute magnitude estimations of annoyance and loudness for 12 recorded HVAC noises containing low frequencies (noise levels in the range 72.8-92 dB SPL and 40.6-61.4 dBA) and found that loudness and annoyance were closely connected, although not always with the same magnitude. The noises that produced a higher rating of annoyance were characterised by dominant low frequency peaks, and the results was suggested by the authors to support previous findings of perceived roughness for such noise. The low frequency noise used in this thesis was of a similar character as the noise used by Hellman and Broner [1999]. It had a dominant content of energy in the centre frequency of 31.5 Hz, although the noise was not judged to have a tonal character. In comparison to the results found by Hellman and Broner [1999], the presence of a tonal character in the low frequency range has been shown to be of little or no importance for annoyance, reduced wakefulness or performance [Holmberg et al. 1993; Landström et al. 1991; Landström et al. 1995]. As mentioned earlier, the guideline for occupational environments in Sweden is given in Aweighted sound pressure levels or, in the case of infrasound, in third-octave band sound pressure levels [AFS 1992:10]. For low frequency noise in the general environment, a guideline with specific recommendations for low frequency noise based on third-octave band analysis [SOSFS 1996:7] could be used. Compared to the latter guideline, a guideline with an increase of 5 dB SPL for each third-octave band has been proposed for occupational environments, at least for work environments where the normal noise level is in the range of 40-50 dB LAeq 8h [Persson Waye 2002]. The suggested guideline comprises only the frequency range of 25 to 200 Hz, and, in order also to cover the higher frequency range, it is therefore necessary to measure the A-weighted sound pressure level. A better guideline ought therefore to cover the whole perceivable frequency range. However, Genuit [1999] pointed out that measurements of third-octave spectrum do not either represent the definitive solution. Genuit [1999] compared three noises with the same third-octave spectrum and sound pressure level and pointed out that, as these two measurements do not take time structures into consideration, the noises can still be annoying to different extents. 75

Taken together, there is an increasing awareness of the importance of sound characteristics in low frequency noise. Besides the agreement that an A-weighted sound pressure level is an inadequate descriptor, there is however still no consensus on how to best describe a low frequency noise. Some authors suggest third-octave band measurements, others loudness, roughness or additional sound characteristics. This thesis evaluated modulation frequency and frequency balance and found that both were of importance for the perceived pleasantness. Further investigations are obviously needed before the best way to describe a low frequency noise to predict adverse effects on performance and annoyance can be confirmed.

Continuation of the work in this thesis - further studies

In compiling the results of this thesis, thoughts about studies to continue the work of this thesis were outlined. It would be interesting to carry out a larger general survey investigating a wider range of different occupational environments, such as lecture rooms, assembly and meeting halls, as other workplaces besides control rooms and office areas can be hypothesised to comprise low frequency noise. Furthermore, it seems motivated to link the home environment to the working environment. For example, how well can a person perform at work if rest and sleep has been disturbed by exposure to low frequency noise or, the other way around, how well does a person sleep after a day of work during exposure to low frequency noise? It would also be interesting to test a wider range of sound characteristics in other low frequency noises in order to get a more complete comprehension of what parameters are perceived as unpleasant. This is needed to be able to make more far-reaching conclusions and to improve present guidelines and recommendations.

76

CONCLUSIONS

Low frequency noise impaired performance on tasks with high and moderate demands on cognitive processing when performed under high workload. Furthermore, low frequency noise impaired performance on some of the low demand tasks and a moderately demanding verbal tasks under low workload. Low demand tasks performed under high workload and tasks evaluating motivation performed under low workload were not affected. The effects were interpreted as being in accordance with the information overload theory and the arousal theory. Low frequency noise was rated as causing a greater impairment of the work capacity and as being more annoying than the reference noise, however the difference between noise conditions was only significant under high workload. Annoyance due to noise was influenced by exposure time and/or sound pressure level, while the rating of impairment of the work capacity primarily was influenced by the kind of noise exposure.

Working under high workload in low frequency noise resulted in elevated cortisol levels among subjects high-sensitive to noise in general, and a tendency towards the same result was found for subjects high-sensitive to low frequency noise. The cortisol response seems to be part of a larger stress pattern involving subjective effects, but no clear relationships between cortisol levels and subjective stress or other subjective reports were found. Adverse effects from low frequency noise during work was most pronounced for subjects high-sensitive to low frequency noise and subjects high-sensitive to noise in general, and the categorisation of sensitivity to low frequency noise could be distinguished from the categorisation of sensitivity to noise in general. Sensitivity to low frequency noise was the individual factor that was concidered to be of greatest importance for negative effects of low frequency noise during work. A more pleasant low frequency noise should contain no or very low modulation frequency and a lower content of frequencies below 500 Hz.

77

ACKNOWLEDGEMENTS

Many thanks to my colleagues at the Department of Environmental Medicine, Göteborg University, for the friendly atmosphere they create, both when we’re working and otherwise and not least for our endless discussions about just about everything. I would like especially to acknowledge the following people: Kerstin Persson Waye, my supervisor, for her guidance, enthusiasm and special support when the going was really tough, by sending encouraging e-mails from the other side of the Kattegat.

Agneta Agge, for her excellent research assistance, her enjoyable company during the laboratory experiments and for always being there when I needed help. Martin Björkman, for his technical assistance and unlimited patience as he taught me time and again how to manage the instruments. Annbritt Skånberg and Cecilia Modigh, for being my sounding board and/or wailing wall, depending on the situation. Ragnar Rylander, for his valuable comments thoughout my research studies and for always making time for me. Yvonne Peterson, for her professional assistance and editing in the preparation of proceedings, articles and, finally, this thesis. The PhD students and senior researchers in the ”Noise Network”, for generously sharing their knowledge, giving me their constructive criticism and being enjoyable company. Jonas Ryberg, for always believing that I could do it (and for saying that I was the best when I felt like I was the worst).

78

This project was funded by the Network for Noise Effects in Working Life at the National Institute for Working Life, Solna, Sweden, and the Swedish Council for Work Life Research, Stockholm, Sweden.

79

REFERENCES

Aardal E, Holm AC. Cortisol in saliva-reference ranges and relation to cortisol in serum. Eur J Clin Chem Biochem 1995;33:927-932. AFS 1992:10 Ordinance on noise. Statute book of the Swedish national board of occupational safety and health, Stockholm, Sweden. Andren L, Lindstedt G, Björkman M, Borg KO, Hansson L. Effect of noise on blood pressure and “stress” hormones. Clin Sci 1982; 62:137-141. Arvidsson O. Doctoral Thesis, On annoyance caused by noise exposure – A relevance study. 1975. SNV PM 590, Department of Hygiene, Karolinska Institute, Stockholm, Sweden. Baddeley AD. A 3 min reasoning task based on grammatical transformation. Psychol Sci 1968;10:341-342. Belojević G, Öhrström E, Rylander R. Effects of noise on mental performance with regard to subjective noise sensitivity. Int Arch Occup Environ Health 1992;64:293-301. Belojević G, Jakovljevic B. Factors influencing subjective noise sensitivity in an urban population. Noise & Health 2001;13:17-24. Bengtsson J, Persson Waye K. Assessments of low frequency noise complaints – and a follow-up study 14 years later. J Low Freq Noise Vibr 2003 (in press). Benton S, Leventhall HG. Experiments into the impact of low level, low frequency noise upon human behaviour. J Low Freq Noise Vibr 1986;5:143-162. Benton S, Leventhall HG. The role of “background stressors” in formation of annoyance and stress responses. J Low Freq Noise Vibr 1994;13:3:95-101.

80

Benton S, Robinson G. The effects of noise on text problem solving for the word processor user (WPU). In: M. Vallet M (ed), Noise and man 1993. Proceedings of the 6th International Congress on Noise as a Public Health Problem, Nice, France, July 1993, pp 539-541. Benton S. Measurement challenges in assessing the annoying characteristics of noise. Is low frequency noise a special case? J Low Freq Noise Vibr 1997;16:13-24. Berglund B, Hassmén P, Soames Job RF. Health effects induced by low-frequency noise: A critical review. Arch Center Sensory Res 1994;1:1-28. Berglund B, Lindvall T (eds). Community Noise. Document prepared for World Health Organization (WHO). Archives of the Center for Sensory Research, Stockholm 1995;2:1-195. Berglund B, Lindvall T, Schwela D, Goh K-T (eds). Guidelines for Community Noise. Document prepared for World Health Organization (WHO) 2000. ISBN: 9971-88-770-3. Bohnen N, Houx P, Nicolson N, Jolles J. Cortisol reactivity and cognitive performance in a continous mental task paradigm. Biol Psychol 1990;31:107-116. Boman E, Hygge S. After effects of noise predictability and task load on motivation. In: D. Cassereau (ed), Proceedings of Inter Noise, Nice, France, 27-30 August 2000, pp 2797-2800. Born J, Hansen K, Marshall L, Molle M, Fehm HL. Timing the end of nocturnal sleep. Nature 1999;387:29-30. Borsky PN. Sonic boom exposure effects 11.4: Annoyance reactions. J Sound Vib 1972;20:477-544. Broadbent DE. A mechanical model for human attention and immediate memory. Psychol Rev 1957;64:205-215.

81

Broadbent DE. Decision and Stress. Academic Press Inc (London) LDT, 1971. Broadhurst PL. The interaction of task difficulty and motivation: the Yerkes-Dodson law revided. Acta psychol 1959;16:321-338. Broner N. Determination of the relationship between low frequency HVAC noise and comfort in occupied spaces – objective phase. ASHRAE 714-RP, Vipac Report 38114, 1994. Bryan ME. Low frequency noise annoyance. In: W. Tempest (ed) Infrasound and Low Frequency Vibration. Academic Press, London, UK, 1976, pp 65-96. Brüel & Kjaer Magazine. Investigators prepares to rocket into space. www.bksv.com. No 2, 2000. Buijis RM, Wortel J, Heerikhuize JJ, Feentra MGP, Ter Horst GJ, Romijin HJ, Falsbeek A. Anatomical and functional demonstration of multisynaptic suprachiasmatic nucleus adrenal (cortex) pathway. Euro J Neurosci 1999:11;1535-1544. Byström M, Landström U, Kjellberg A. Effects of tones and brodband noise on the perception of annoyance during work with different tasks (in Swedish). Arbete och Hälsa, 1991, National Institute for Working Life, Solna, Sweden. Cohen S. After effects of stress on human performance and social behaviour: A review of research and theory. Psychol Bullentin 1980; 88:1:82-108. Cohen S, Evans GW, Stokols D, Krantz DS. Behaviour, Health, and Environmental Stress. Plenum Press. New York and London, 1986. Dornic S. Noise and information processing: findings, trends and issues. Report 715, 1990, Dept of Psychology, University of Westminister, London, UK. Doyle A, Hucklebridge E, Evans P, Clow A. Salivary monoamine oxidase A and B inhibitory activities correlate with stress. Life Sci 1996;59:1357-1362.

82

Evans G, Johnson D. Stress and open-office noise. J Appl Psychol 2000;85:5:779-783. Evans GW. Motivational consequences of exposure to noise. In: N. Carter, R.F. Soames Job (eds) Proceedings from the 7th International Congress on Noise as a Public Health Problem, Sydney, Australia, 22-26 November 1998, pp 311-320. Evans P, Bristow M, Hucklebridge F, Clow A, Pang F-Y. Stress, arousal, cortisol and secretory immunoglobin A in students undergoing assessment, Br J Clin Psychol 1994;33:575-576. Frankenhauser M, Lundberg U. Immediate and delayed effects of noise on performance and arousal. Biol Psychol 1974;2:127-133. Fruhstorpher B, Pritsch MG, Ott P. Effects of daytime noise load on the sleep-wake cycle and endocrine patterns in man: II. 24 hours secretion of anterior and posterior pituitary hormones and of cortisol. Intern J Neurosci 1988;39:211-221. Gamberale F, Iregren A, Kjellberg A. SPES: The computerized Swedish Performance Evaluation System. Background, critical issues, empirical data and users manual. National Board of Occupational Safety and Health, Solna, Sweden, 1989. Genuit K. The use of psychoacoustic parameters combined with A-weighted SPL in noise description. In: Proceedings of the 1999 International Congress and Exposition on Noise Control and Engineering, December 6-8, Fort Lauderdale, Florida, USA, pp 1887-1892, 1999. Glass DC, Singer JE. Urban stress: Experiments on noise and social stressors. Academic Press, Inc, 111 Fifth Avenue, New York, USA, 1972, pp 3-182. Goldstein M, Kjellberg A. Annoyance and low frequency noise with different slopes of frequency spectrum. J Low Freq Noise Vibr 1985;2:43-51. Guski R. Personal and social variables as co-determinants of noise annoyance. Noise & Health 1999;3:45-56.

83

Guski R, Felscher-Suhr U, Schuemer R. The concept of noise annoyance: How international experts see it. J Sound Vib 1999;223:513-527. Hallmann S, Guski R, Schuemer R. What residents have in mind when asked for scaling noise annoyance. In: A. Selamet, R. Singh, G.C. Maling (eds) Proceedings of the 2002 International Congress and Exposition on Noise Control and Engineering, August 22, Dearborn, Mi, USA, 2002, N173. Hebb DO. Drives and the C:N:S (conceptual nervous system). Psychol Rev 1955;62:243-254. Hellmann R, Broner N. Assessment of the loudness and annoyance of low.frequency noise from a psychologial perspective. In: Proceedings of the 1999 International Congress and Exposition on Noise Control and Engineering, December 6-8 1999, Fort Lauderdale, Florida, USA, pp 1893-1898, 1999. Hockey GRJ. Changes in operator efficiency as a function of environmental stress, fatigue, and circadian rhythms. In: K.R. Boff , L. Kaufman, J.P. Thomas (eds). Handbook of Perception and Performance. Vol. II. Cognitive process and performance, Wiley, New York, 1986, chapter 44, pp 1-49. Hockey GRJ. Stress and cognitive components of skilled performance, In: V. Hamilton, D. Warburton (Eds), Human Stress and Cognition, Wiley, New York, 1979. Holmberg K, Landström U, Kjellberg A. Effects of ventilations noise due to frequency characteristics and sound level. J Low Freq Noise Vibr 1993;12:115-122. ISO 226. Normal equal-loudness level. International organiziation for standization, Geneva, 1987. ISO 389. Acoustic reference zero for the calibration of audiometric equipment. Part 7: Reference threshold of hearing under free-field and diffuse-field listening conditions, 1996. Ising H, Braun C. Acute and chronic endocrine effects of noise: Review of the research conducted at the Institute of water, soil and air hygiene. Noise & Health 2000;7:7-24. 84

Ising H, Ising M. Chronic cortisol increases in the first half of the night caused by road traffic noise. Noise & Health 2002;7:13-21. Jelinkova Z. Coping with noise in noise sensitive subjects. In: B. Berglund, U. Berglund, J. Karlsson, T. Lindvall (eds) Proceedings of the 5th International Congress on Noise as a Public Health problem, Stockholm, Sweden, August 21-25, 1988, pp 27-30. Job RFS. Noise sensitivity as a factor influencing human reaction to noise. Noise & Health 1999;3:57-68. Jones DM, Miles C, Page J. Disruption of proofreading by irrelevant speech: Effects of attention, arousal or memory? Appl Cognitive Physiol 1990;4:89-108. Jones DM, Davies DR. Individual and group differences in the response to noise. In: D. Jones, A.J. Chapman (eds) Noise and Society, John Wiley & Sons Ltd, 1984, pp 125-153. Kahneman D. Attention and Effort. Prentice-Hall Inc., Englewood Cliffs, New Jersey, 1973. Key KF, Payne MC. Effects of noise on performance and annoyance for women and men. Perceptual and Motor Skills 1981;52:435-441. Kirschbaum C, Hellhammer DH. Salivary cortisol in psychoneuroendocrine research: Recent developments and applications. Psychoneuroendocrinology 1994;19:313-333. Kirschbaum C, Pruessner JC, Stone AA, Federenko I, Gaab J, Lintz D, Scommer N, Hellhammer DH. Persistent high cortisol responses to repeated psychological stress in a subpopulation of healthy men. Psychoses Med. 1995;57:468-474. Kirschbaum C, Wolf OT, May M, Wippich W, Hellhammer DH. Stress- and treatmentinduced elevations of salivary cortisol associated with impaired declarative memory in healthy adults. Life Sci 1996;58:1475-1483. Kirschbaum C, Gonzalez-Bono E, Rohleder N, Gessner C, Pirke KM, Salvador A, 85

Hellhammer DH. Effects of fasting and glucose load on free cortisol responses to stress and nicotine. J Clin Endocrinol Metab 1997;82:1101-1105. Kirschbaum C, Hellhammer D. Salivary cortisol, Encyclopedia of stress, 2000;3:379-383. Kjellberg A. Effects of sleep deprivation on performance of a problem solving task. Psychol Reports 1975;37:479-485. Kjellberg A. Sleep deprivation and some effects on performance. I. Problems of arousal change. II. Lapses and other attentional effects. III. Motivation, comments and conclusions. Waking Sleeping, 1977;1:139-153. Kjellberg A, Goldstein M, Gamberale F. An assessment of dB(A) for predicting loudness and annoyance of noise containing low frequency components. J Low Freq Noise Vibr 1984;3:10-16. Kjellberg A, Goldstein M. Loudness assessment of band noise of varying bandwidth and spectral shape. An evaluation of various frequency networks. J Low Freq Noise Vibr 1985;4:12-26. Kjellberg A, Wide P. Effects of simulated ventilation noise on performance of a grammatical reasoning task. In: B. Berglund, U. Berglund, J. Karlsson, T. Lindvall (eds) Proceedings of the 5th International Congress on Noise as a Public Health problem, Stockholm, Sweden, August 21-25, 1988, pp 31-36.

Kjellberg A, Iwanowski S. The stress/energy questionnaire – Development of a method to evaluate estimations of mood at work. Report 1989:26 (in Swedish). National Institute of Occupational Health, Solna, Sweden. Kjellberg A, Landström U. Noise in the office: Part II - The scientific basis (knowledge base) for the guide. Int J Ind Ergonomics 1994;14:93-118. Kjellberg A, Muhr P, Sköldström B. Fatigue after work in noise – an epidemiological

86

survey study and three quasi-experimental field studies. Noise & Health 1998;1:47-55. Kudielka BM, Kirschbaum C. Awakening cortisol response are influenced by health status and awakening time but not by menstrual cycle phase. Psychoneuroendocrinology 2003;28:35-37. Kugler J, Reintjes F, Tewes V, Schedlowski M. Competition stress in soccer coaches increases salivary Immunoglobulin A and salivary cortisol concentrations. J Sport Med Phys Fitness 1996;36:117-120. Kragh J, Jonasson H, Lyse NHl. Road Traffic Noise - Nordic Prediction Method. 1996, Nordic Council of Ministers. TemaNord 1996;525:1-36. Landström U, Liszka L, Danielsson Å, Linsmark A, Lindqwist M, Söderberg L. Changes of wakefulness during exposure to infrasound. J Low Freq Noise Vibr, 1982;1:2:79-87. Landström U, Lundström R, Byström M. Exposure to infrasond - perception and changes in wakefulness. J Low Freq Noise Vibr, 1983;2:1-11. Landström U. Laboratory and field studies on infrasound and its effects on humnas. J Low Freq Noise Vibr 1987;6:29-33. Landström, U. Noise and fatigue in working environments. Environ Int 1990;16:471-476. Landström U, Kjellberg A, Söderberg L. Spectral character, exposure levels and adverse effects of ventilation noises in offices. J Low Freq Noise Vibr 1991;10:83-91. Landström U, Kjellberg A, Byström M. Acceptable levels of sounds with different spectral characteristics during the performance of a simple and a complex non-auditory task. J Sound Vib 1993;160:533-542. Landström U, Åkerlund E, Kjellberg A, Tezars M. Exposure levels, tonal component and noise annoyance in working environment. Environ Int 1995;21: 265-275.

87

Landström U, Byström M, Kjellberg A, Nordström B. Annoyance during exposure to amplitude modulated noise. Report 1996:16 (in Swedish). National Institute for Working Life, Solna, Sweden. Landström U, Byström M, Kjellberg A, Nordström B. (a) Performance and annoyance during exposure for amplitude-modulated noise at different noise levels. Report 1997:15 (in Swedish). National Institute of Occupational Health, Solna, Sweden. Landström U, Kjellberg A, and Söderberg L. (b) Noise annoyance at different times of the working day. In: Proceedings of the 8th International conference on low frequency noise and vibration, Gothenburg, Sweden 3-5 June, 1997, pp 53-61. Landström U, Arlinger S, Hygge S, Johansson Ö, Kjellberg A, Persson Waye K. Annoying noise. Review for criteria documentation (in Swedish). Report 1999:7. National Institute for Working Life, Solna, Sweden. Lundberg U. Coping with stress: neuroendocrine reactions and implications for health. Noise & Health 1999;4:67-74. Levi L. Stress – a review. International health perspective. In: R. Ekman, B. Arnetz (eds) Stress, 2002, pp 44-61 (in Swedish). Liber AB, 113 98 Stockholm, Sweden. Melamed S, Bruhis S. The effects of chronic industrial noise exposure on urinary cortisol, fatigue and irritability- A controlled field experiment. J Occup Environ Med 1996;38:252256. Miller WL, Tyrell JB. The adrenal cortex. In: P. Fetig, J.D. Baxter, L.A. Frohman (eds), Endocrinology and Metabolism. New York: McGraw Hill Inc., 1995, pp 555-711. Miki K, Kawamorita K, Araga Y, Musha T, Sudo A. Urinary and salivary stress hormone levels while performing arithmetic calculation in noisy environment. Ind Health 1998;36:6669. Mirowska M. An investigation and assessment of annoyance of low frequency noise in 88

dwellings. J Low Freq Noise Vibr 1998;17:119-126. Nagai N, Matsumoto M, Yamasumi Y, Shiraishi T, Nishimura K, Matsumoto K, Miyashita K, Takeda S. Process and emergence on the effects of infrasonic and low frequency noise on inhabitants. J Low Freq Noise Vibr 1989;8: 87-99. Nightingale F. Noise. In: M. Skeet (ed) Notes on Nursing. Pp 59-69, 1980. Churchill Livingstone, Edinburgh. Ortiz G A, Argüelles HA, Sposari G, Villafaňe CT. Modification of epinephrine, norepinephrine, blood lipid fractions and the cardiovascular system produced by noise in an industrial medium. Hormone Res 1974;5:57-64. Osguthorpe DJ, Mills JH. Nonauditory effects of low-frequency noise exposure in humans. Otolaryngology – Head and Neck Surgery 1982;90:367-370. Pawlaczyk-Luszcynska M, Dudarewicz A, Waszkowska M. Annoyance of low frequency noise in control rooms. In: A. Selamet, R. Singh, G.C. Maling (eds) Proceedings of the International Congress and Exposition on Noise Control and Engineering, August, Dearborn, Mi, USA, 2002, N118. Persson K, Rylander R. Disturbance from low-frequency noise in the environment: A survey among the local environmental health authorities in Sweden. J Sound Vib 1988;121:339-345. Persson K, Björkman M, Rylander R. Loudness, annoyance and dBA in evaluating low frequency sounds. J Low Freq Noise Vibr 1990;9:32-45. Persson K, Björkman M, Rylander R, Hellström P-A. A pilot study evaluating effects on vestibularis and subjective symptoms among sensitive subjects exposed to low frequency sounds. In: Proceedings of the 7th international meeting on low frequency noise and vibration. Edinburgh 1993, pp 35-140. Persson Waye K. Doctoral Thesis, On the effects of environmental low frequency noise, 1995. Department of Environmental Medicine, Göteborg University, ISBN 91-628-1516-4. 89

Persson Waye K, Rylander R, Benton S, Leventhall HG. Effects on performance and work quality due to low frequency ventilation noise. J Sound Vib 1997;205:467-474. Persson Waye K, Rylander R. The prevalence of annoyance and effects after long-term exposure to low frequency noise. J Sound Vib 2001;240:483-497. Persson Waye K. Effects of low frequency noise in the occupational envirornment – present knowledge base. In: A. Selamet, R. Singh, G.C. Maling (eds) Proceedings of the International Congress and Exposition on Noise Control and Engineering, August, Dearborn, Mi, USA, 2002, N273. Persson Waye K, Clow A, Edwards S, Hucklebridge F, Rylander R. Effects of low frequency noise on the cortisol response to awakening and sibjective sleep quality. Life Sci 2003;72:863-875. Peter JH, Cassel W, Ehrig B, Faust M, Fuchs E, Langanke P, Meinzer K, Pfaff U. Occupational performance of a paced secondary task under conditions of sensory deprivation. Eur J Appl Physiol, 1990;60:309-314. Plihal W, Krug R, Pietrowski R, Fehm HL, Born J. Corticosteroid receptor mediated effects on mood in humans. Psychoneuroendocrinology 1996;21:515-523. Poulsen T. Influence of session length on judged annoyance. J Sound Vib 1991;145:217224. Preis A. Measurement of annoyance components. In: F. Alison Hill, R. Lawrence (eds) Proceedings of Inter Noise, Liverpool, England, 1996, pp 2289-2294. Pruessner JC, Wolf OT, Hellhammer DH, Buske-Kirschbaum K, von Auer S, Jobst F, Kaspers F, Kirschbaum C. Free cortisol levels after awakening: a reliable biological marker for the assessment of adrenocortical activity. Life Sci 1997;61:2539-2549. Pruessner JC, Hellhammer DH, Kirschbaum C. Burnout, perceived stress, and cortisol 90

responses to awakening. Psychosomatic Med 1999; 61: 197-204. Rylander R, Persson Waye K. Human perception of environmental noise in acute and chronic exposure conditions. In: F. Augusztinovicz (ed) Proceedings of Inter Noise, Budapest, Hungary 25-27 August, 1997, pp 971-976. Schultz P, Merck D. Gesundheitsrelevante korrelate eines erniedringen cortisolspiegels: Eine undersuchung an angestellten im dentaltechnichen labor. Zeitschr Gesundheitspsychol 1997; 14: 47-62. Schultz P, Kirschbaum C, Prüssner J, Hellhammer D. Increased free cortisol secretion after awakening in chronically stressed individuals due to work overload. Stress Med 1998;14:91-97.

Seligman MEP. Helplessness, 1975. W.H. Freeman, San Francisco, USA. Sjöberg L, Svensson E, Persson LO. The measurement of mood. Scand J Psychol 1979;20:1-18. Smith AP, Broadbent DE. Noise and levels of processing. Acta Psychologica 1981;47:129142. Smith AP, Miles C. The combined effects of occupational health hazards: an experimental investigation of the effects of noise, night work and meals. Int Arch Occup Environ Health 1987;59:83-89. Smith AP, Jones DM. Noise and performance. In: D.M. Jones, A.P. Smith (eds). Handbook of Human Performance, Vol 1: The Physical Environment. Academic Press, London, 1992, pp 1-28. Smith A, Hayward S, Rich N. Perceptions of aircraft noise exposure, noise sensitivity, sleep disturbance and health: Results of the Bristol noise, sleep and health study. In: D. Cassereau (ed) Proceedings of Inter Noise, Nice, France 2000, pp 984-987.

91

Smith A, Rich N. Helicopter noise, noise sensitivity, annoyance and cardiovascular response. In: A. Selamet, R. Singh, G.C. Maling (eds) Proceedings of the International Congress and Exposition on Noise Control and Engineering, August, Dearborn, Mi, USA, 2002, N532. SOSFS 1996:7/E. Indoor Noise and High Sound-Levels. General Guidelines issued by the Swedish National Board of Health and Welfare, 1996. Smyth J, Ockenfels M C, Porter L, Kirschbaum C, Hellhammer DH, Stone AS. Stressors and mood measured on a momentary basis are associated with salivary cortisol secretion. Psychoneuroendocrinology 1998;23:353-370. Stansfeld SA. Noise, noise sensitivity and psychiatric disorder. Psychol Med 1992;22: 1-44. Tafalla R, Evans GW. Noise, physiology, and human performance: The potential role of effort. J Occup Health Psychol 1997;2:148-155. Tang SK, Burnett J, Poon CM. Aural environment survey in air-conditioned open-plan offices. Proc. CIBSE A. Building Serv Eng Res Technol 1996:17:2:97-100. Tesarz M, Kjellberg A, Landström U, Holmberg K. Subjective response patterns related to low frequency noise. J Low Freq Noise Vibr Active Contr 1997;16:2: 145-149. Tokita Y. Low frequency noise pollution problems in Japan. In: H. Møller, P. Rubak (eds) Proceedings of the Conference Low frequency noise and hearing, Aalborg, Denmark, 1980, pp 189-196. Topf M. Stress effects of personal control over hospital noise. Behavioral medicine: an interdisciplinary journal of research and practice 1992;2:84-94. Von Muggenthaler E. Low frequency and infrasonic vocalizations from tigers. www.animalvoice.com. Fauna communication research institute, North Carolina, USA. Weinstein ND. Effects of noise on intellectual performance. J Appl Psychol 1974;59:54892

554. Weinstein ND. Individual differences in reactions to noise: A longitudinal study in a college dormitory. J Appl Psychol 1978;63: 458-466. Weinstein ND. Individual differences in critical tendencies and noise annoyance. J Sound Vib 1980;68):241-248. Yamada S, Watanabe T, Kosaka T, Negishi H, Watanabe H. Physiological effects of low frequency noise. J Low Freq Noise Vibr 1986;5:14-25. Yamamura K, Maehara N, Sadmoto T, Harabuchi. Effect of intermittent (traffic) noise on man: temporary threshold shift, and change in urinary 17-OHCS and salivary cortisol levels, Eur J Appl Physiol 1982:48:303-314. Yerkes RM, Dodson JD. The relation of strength of stimulus to rapidity of habit-formation. J Comp Neurol Psychol 1908;18:459-482. Zeier H, Brauchli P, Joller-Jemelka HI. Effects of work demand on immunoglobulin A and cortisol in air traffic controllers. Biol Psychol 1996;42: 413-423. Zimmer K, Ellermeier W. Psychometric properties of four measures of noise sensitivity: a comparison. J Environ Psychol 1999;19:295-302. Zwicker E, Fastl H. Psychoacoustics. Facts and models. Second edition, Springer-Verlag, Heidelberg, Germany, 1999. Öhrström E, Björkman M, Rylander R. Noise annoyance with regard to neurophysiological sensitivity, subjective noise, noise sensitivity and personality variables. Psychol Med 1988;18: 605-613.

93