Road and rail traffic noise induce comparable extra-aural effects as revealed during a short-term memory test.

To examine extraaural effects as induced by 20 min of road (ROAD) and 20 min of rail (RAIL) traffic noise with same loudness (75 dBA), a laboratory study was carried out. The study (N = 54) consisted of 28 high and 26 low-annoyed healthy individuals as determined by a traffic annoyance test. To control attention, all individuals performed a nonauditory short-term memory test during the noise exposures. A within-subject design, with phases of ROAD, RAIL, and CALM (memory test only), alternated by phases of rest, was defined. Heart rate (HR), systolic blood pressure (sBP), total peripheral resistance (TPR), as well as three autonomic variables, preejection period (PEP), 0.15-0.4 Hz high-frequency component of HR variability (HF), and salivary stress biomarker alpha amylase (sAA) were measured. In relation to CALM, HR increased (RAIL +2.1%, ROAD +2.5%), sBP tended to increase against the end of noise exposure, PEP decreased (RAIL -0.7%, ROAD -0.8%), HF decreased (RAIL -3.4%, ROAD -2.9%), and sAA increased (RAIL +78%, ROAD +69%). No differences were found between RAIL and ROAD, indicating that both noise stressors induced comparable extraaural effects. Factor annoyance showed significant during CALM. Here a reduced sympathetic drive (higher PEP values) combined with an increased vascular tone (higher TPR values) was found at the high-annoyed subgroup.

INTRODUCTION

Exposure to noise in the environment from transport sources is an increasingly prominent feature. The direct effects of sound energy on human hearing are well established and accepted. Traffic noise is commonly presented at a level clearly below the noise level causing hearing damage, so that aural effects can be neglected. In contrast, nonauditory effects of noise on human health are not the direct result of sound energy. Instead, such extraaural effects are the result of noise as a general stressor. Extraaural effects may include annoyance, mental health, sleep disturbance, and alteration of physiological functions.

It is well accepted that traffic noise of moderate sound pressure level (70–80 dBA) induces noticeable cardiovascular alterations.[12] In healthy individuals, when exposed to experimental noise, such alterations are shown by accelerated heart rate (HR), elevated systolic blood pressure (sBP), and increased peripheral vascular resistance, see Griefahn[3] and Stansfeld and Matheson[4] for review. Basically such alterations correspond to changes in the coactivation of the two branches of the autonomic nervous system (ANS). As parasympathetic activity dominates in the resting state, quick withdrawal of this activity is expected in the early phase of noise exposure. Furthermore, with ongoing noise exposure, increase of sympathetic activity is expected, including elevated release of stress biomarkers via the sympathetic-adreno-medullary (SAM) axis.

For observing such stress responses indirectly, frequency components of heart rate variability (HRV) have been described.[5] The main components are the 0.15–0.4 Hz high-frequency (HF) band indicating to parasympathetic activity, and the 0.04–0.15 Hz low-frequency (LF) band indicating both parasympathetic and sympathetic activities.[6] As a specific indicator for sympathetic activity, isovolumetric contractile phase of ventricular systole, also known as preejection period (PEP), was proposed[7] and employed to stress studies.[8] Further testing of salivary biomarkers has become widely accepted to assess the main axis of stress, the hypothalamic-pituitary-adrenal axis and the SAM axis.[9] The salivary stress biomarker alpha amylase (sAA) reflects the activity of the SAM axis, as reported in several experimental studies,[101112] showing an increased secretion of sAA under different stress conditions including exposure to traffic noise.

The main focus of the current research is the comparison of extraaural effects as induced by 20 min of naturalistic road (ROAD) and 20 min of naturalistic rail traffic noise (RAIL). Although both noise stressors were precisely adjusted to obtain the same long-term A-weighted energy equivalent sound level (LEQ) of 75 dBA, the time structure of ROAD and RAIL is different. RAIL shows lesser events but induces higher peak sound pressure levels than ROAD [see Figure 1]. Therefore, it appeared reasonable that RAIL induces stronger effects, which is the working hypothesis of this research. In addition, as the induced effects may depend on the disposition to traffic noise[13] it appeared reasonable to examine a collective group consisting of low and high-annoyed individuals.

Figure 1

Instantaneous sound pressure level of rail (upper plot) and road traffic noise (lower plot) as presented in this study. For both noises, the long-term equivalent continuous sound level (LEQ) over the time interval of 1200 s was 75 dBA

To test this hypothesis with normal individuals, a randomized within-subject test protocol was developed. This protocol includes a nonauditory cognitive task (short-term memory test) during the 20 min noise exposures to control the individual's attention and avoid drowsiness and mind wandering while sitting in the arm chair. Cardiovascular assessments were performed with the TASK Force Monitor (CNSystems Medizintechnik AG, www.cnsystems.at), a device allowing extraction of beat-to-beat hemodynamic and autonomic variables from electrophysiological and blood pressure (BP) measurements. The investigated hemodynamic variables are HR, sBP, and total peripheral resistance (TPR), and the autonomic variables are HF for vagal drive and PEP for sympathetic drive. In addition to the cardiovascular assessments, salivary stress biomarker sAA was determined once before and at the end of each noise exposure.

METHODS

Participants

The participants of this study consisted of 59 healthy individuals (30♀, 29♂; aged between 20 and 58 years) with no record of cardiovascular disease. The participants were recruited from a larger database of 200 addresses of individuals who performed a standardized traffic noise annoyance test based on an 11-graded interval scale; for more details, see earlier article by Cik et al.[14] Depending on the test score, participants were assigned to either the high (15♀, 15♂) or the low-annoyed subgroup (♀15, 14♂). All participants lived in quiet residential locations and had no history of occupational noise exposure. To ensure normal hearing, participants were tested by tone audiometry in the frequency band between 200 and 8 kHz. Furthermore, the recorded electrocardiography (ECG) signals and BPs of each participant were diagnosed. The study was approved by the local Ethics Committee of the Medical University of Graz, Graz, Austria. All study participants gave written informed consent.

Naturalistic traffic noise

For the experimental tests, pass-by road and rail traffic vehicle noise, taken from an existing database of the Institute of Highway Engineering and Transport Planning, Graz University of Technology, Graz, Austria, was selected and used for presentation.[13] The road and rail traffic pass-by samples contained A-weighted sound levels with 75 dBA, typical A-weighted frequency spectra meeting the national standards of traffic noise frequencies (OENORM EN 1793-3: 1998 03 01:N). As shown in Figure 1, rail noise (upper plot) exhibits few slowly changing events with peak sound levels of about 90 dBA, whereas road noise (lower plot) exhibits several events with peak levels just below 80 dBA. The sounds were presented in a specially adapted anechoic listening room (noise level <30 dBA) via a stereophonic loudspeaker system.

Physiological recordings and data preprocessing

The cardiovascular assessments included ECG, thoracic impedance cardiography (ICG), and continuous BP, all recorded with the Task Force Monitor (TFM). Briefly, this monitor uses a standard three-electrode system for ECG, a four-electrode system for ICG, and two inflatable finger cuffs in combination with a standard arm cuff for BP. For detection of BP, the vascular unloading principle[15] is utilized. The two cuffs, one for the middle and one for the index finger, are operated alternately to avoid pain and stress. The TFM includes a database for storage, and software for processing of the recorded beat-to-beat data. During the recording sessions, a 2-min clock signal was delivered to the event marker input of the TMF to trigger time stamps in the database.

The TFM further includes the software for calculation of the hemodynamic (HR, sBP, TPR) and the autonomic beat-to-beat values (HF, PEP). For example, TPR is calculated from mean arterial pressure divided by stroke index and HR. Stroke index is obtained by two steps: first an estimate of stoke volume is determined from the ICG and the ECG by a suitable algorithm,[16] and then this value was normalized by an estimate of body surface. PEP is determined by the time interval between the R wave of the ECG and the B point of the ICG.

After the recordings, mean values were calculated over the 2-min intervals as marked by the time stamps. Outliers as produced by maldetections of the TFM software or motion artifacts were removed. Missing data points were estimated from adjacent data points and replaced. The presented power values of HF were log10 transformed to obtain normal distributions.

Testing of sAA

Saliva samples were collected using the absorption-based Salivette® collection device according the manufacturers’ instructions (Sarstedt GmbH, Nümbrecht, Germany). Saliva sampling was done prestress and poststress phases and stored at −70°C until being analyzed. Prior to the analysis, saliva samples were thawed and centrifuged for 10 min at 4000 rpm and 200 μl of each saliva sample was then diluted 1:100 with distilled water. The determination of sAA was performed on a Hitachi 912 laboratory analyzer (Roche Diagnostics, Mannheim, Germany) using the corresponding alpha-amylase kit (Roche Diagnostics). Alpha-amylase activity was expressed as international units per liter of saliva (U/l).

Nonauditory short-term memory task

For control of the participant's attention, a modified variant of Sternberg's item recognition task[17] was programmed on a computer. Each trial started with a plus sign in the middle of the screen. Then a series of six items (letters from the German alphabet, in random order) successively were presented, one item per second. Then after a time delay of 3 s, a seventh item (probe letter) was presented. By mouse clicks (YES or NO), participants had to decide whether the seventh item was included in the preceding series or not.

The total time for one trial was 20 s. The probe letter was part of the preceding series in 50% of the trials. During one noise exposure (20 min), typically 60 decisions from 60 randomized trials were obtained.

Experimental procedure and study design

To minimize influence from diurnal rhythm (especially for determination of sAA), experimental tests were performed between 1:00 to 4:00 pm. Prior to the tests, participants filled out a questionnaire about size, age, and sex, medications currently taken, sleep quality of the last night, and physical or mental stress on this day. Then the participants were informed about the measurements, the experimental procedure, and the short-term memory test. To avoid high stress levels induced by the memory test, participants were instructed not to focus on speed and reaction time during performing this test. Participants performed a 10-min test phase to get acquainted with this test.

The experiment started with fixation of electrodes for ECG and ICG and positioning of finger and arm cuffs on the nondominant hand. Participants were seated on a comfortable armchair with adjustable foot rests, in front of the computer display. Then after checking the quality of the cardiovascular signals on the TFM, participants were asked to relax. Thereafter, the prenoise saliva samples were taken and data recording was started. Duration of data recording was 100 min for one complete experiment. An experiment consisted of four resting periods (REST; each 10 min) lacking any acoustic and mental stress and three active periods (ACTIVE; each 20 min), according to the scheme REST–ACTIVE–REST–ACTIVE–REST–ACTIVE–REST. After each active period, participants were allowed to move in the chair until to the first half of the resting period. Preexperiments demonstrated that after a noise exposure of 20 min, the assessed variables returned to baseline within 10 min.

Memory tests were exclusively performed during the active periods. During these periods, either road noise (ROAD), rail noise (RAIL), or no noise (CALM) was presented. To counterbalance time-order effects (adaptation because of later sound presentations in relation to the first presentation within a track), three sound tracks were produced: RAIL–ROAD–CALM (RRC), ROAD–CALM–RAIL (RCR), and CALM–RAIL–ROAD (CRR). Participants were not informed about their sound track.

Statistical analysis

Physiological data was selected according to the three test conditions RAIL, ROAD, and CALM resulting into nine 2-min mean values (T2, T4, T6, T8, T10, T12, T14, T16, T18) per condition. The 10th value (T20) was omitted because during 18–20 min, the saliva samples were taken. For inspection of the physiological responses [Figure 2], additionally the 2-min mean value before onset of a stressor was selected. For analysis of variance (ANOVA), the nine 2-min mean values (T2, T4, T6, T8, T10, T12, T14, T16, T18) were pooled into one mean value. To determine the level of the cardiovascular variables at the resting state, the three values at REST were pooled into one mean value.

Figure 2

Hemodynamic (upper tree plots) and autonomic cardiovascular variables (lower two plots) in response to RAIL, ROAD, and CALM. Always the first bar is the value at REST, followed by the response values at time levels T2, T4, T6, T8, T10, T12, T14, T16, and T18; mean values with standard error

For statistical analyses, a repeated measures ANOVA with within-subject factor SOUND (RAIL, ROAD, CALM), and with inter-subject factors SEX (male, female), ANNOYED (high, low), and TRACK (RRC, RCR, CRR) was conducted for each variable (HR, sBP, TPR, PEP, HF, sAA). If significant main effects and interactions were indicated by the four-factorial ANOVAs, follow-up ANOVAs were performed if necessary and post-hoc comparisons were conducted. Probability values less than 0.05 were considered as significant, and values less than 0.1 were indicated as a trend. Amplitude values are given by mean ± standard deviation in the text, and by mean ± standard error in the figures.

RESULTS

Data from physiological recordings and saliva sample collection were obtained from all 59 individuals who participated in the study. After inspection of the recorded signals, five participants were excluded from the statistical analysis: one participant because of a pathological morphology in the QRS complex (ECG), three participants because of tachycardia (HR > 100 bpm), and one participant because of arterial hypertension (sBP > 160). Thus, a total of 54 study participants (28 high annoyed, 26 low annoyed) contributed to the results.

Hemodynamic variables

The estimated hemodynamic variables were HR, sBP, and TPR. In Figure 2 (upper three plots), the averaged responses to the three stressors RAIL, ROAD, and CALM in relation to REST are shown. HR at REST was 67.8 ± 9.7 bpm. HR rapidly accelerated after onset the stressors and then remained almost constant over exposure time. sBP at REST was 119.9 ± 11.2 mmHg. sBP progressively elevated after onset of the stressors RAIL and ROAD, whereas for CALM, only a minor effect is observed. TPR at REST was 2548 ± 609 dyne · s · m2/cm5. TPR increased after onset of the stressors RAIL and ROAD, whereas for CALM, only a minor increase is observed. TPR and sBP show considerable variations, obviously because of the recalibrations and switching of the finger cuffs.

For comparisons within states of ACTIVE, the repeated measures ANOVAs give account. The ANOVA for HR revealed main effect on SOUND (F=12.77, P < 0.001) and the first-order interactions with SOUND are: SOUND × ANNOYED (F=2.19, P=0.118), SOUND × TRACK (F=4.03, P=0.005), and SOUND × SEX (F=1.22, P=0.301). Post-hoc comparisons on main effect SOUND revealed accelerated HR for RAIL (+2.1%, P < 0.001) and for ROAD (+2.5%, P < 0.001) in relation to CALM. No significant difference was found between RAIL and ROAD. Follow-up ANOVAs were conducted with factor SOUND, separately for each TRACK: RRC (F=12.38, P < 0.001), RCR (F=8.33, P=0.001), and CRR (F=1.48, P=0.085). Post-hoc comparisons for TRACK RRC revealed higher HR for RAIL (71.7 ± 8.2 ppm, P < 0.001) and for ROAD (71.4 ± 7.0 ppm, P < 0.001), both in relation to CALM (69.0 ± 6.8 ppm). Post-hoc comparisons on TRACK RCR revealed higher HR for ROAD (70.9 ± 9.7 ppm) in relation to RAIL (68.9 ± 9.5 ppm, P=0.028) and in relation to CALM (68.3 ± 9.4 ppm, P > 0.001). No significant differences were found for TRACK CRR.

The ANOVA for sBP revealed main effect on SEX (F=10.85, P=0.002), whereas there was no main effect on SOUND (F=2.12, P=0.115). Post-hoc comparison with inter-subject factor SEX (pooled data) revealed elevated sBP (P < 0.001) for men (125.8 mmHg) in relation to women (117.3 mmHg). As sBP during RAIL and ROAD slightly elevated during noise exposure [Figure 2], paired samples t tests were performed with pooled data from the first (T2, T4, T6, T8) and from the second half (T12, T14, T16, T18) of exposure time. Here no differences were found at the first half of exposure time if compared to CALM. However, at the second half of exposure time, an elevation was found for RAIL when compared with CALM (+1.7%, P=0.014). No significant differences were found between RAIL and ROAD.

The ANOVA for TPR revealed main effects on ANNOYED (F=4.41, P=0.043) and SEX (F=5.22, P=0.027). No main effect was revealed for SOUND (F=3.00, P=0.055). Post-hoc paired comparisons with inter-subject factor ANNOYED revealed higher TPR values for the high-annoyed subgroup at CALM (+13.8%, P=0.049), and trends to higher values at RAIL (+12.1%, P=0.069) and ROAD (+12.8%, P=0.077). Post-hoc paired comparisons with inter-subject factor SEX revealed higher TPR values for man in relation to women at RAIL (+13.8%, P=0.039), ROAD (14.9%, P=0.038), and CALM (+17.6%, P=0.013).

Autonomic cardiovascular variables

The estimated autonomic variables were PEP and HF. Figure 2 (lower two plots) shows the averaged responses to the stressors RAIL, ROAD, and CALM in relation to REST. PEP at REST was 125.6 ± 5.9 ms. PEP quickly decreased after onset of the stressors (corresponding to an increase of sympathetic activity) and then remained almost constant over exposure time. HF at REST was 6.28 ± 0.80. HF decreased (corresponding to a withdrawal of parasympathetic activity) and then remained constant too. Both autonomic variables showed higher response amplitudes to stressors RAIL and ROAD and then to CALM.

The repeated measures ANOVA for PEP revealed main effects on SOUND (F=5.30, P=0.007) and ANNOYED (F=5.11, P=0.029), and demonstrated interaction TRACK × SEX (F=4.36, P=0.019). The first-order interactions with SOUND are: SOUND × ANNOYED (F=2.34, P=0.102), SOUND × TRACK (F=1.32, P=0.271), and SOUND × SEX (F=0.71, P=0.493). Post-hoc comparisons on factor SOUND revealed a decrease in PEP for RAIL (−0.7%, P=0.022) and ROAD (−0.8%, P < 0.001), both in relation to CALM. No significant difference was found between RAIL and ROAD. Post-hoc comparison on inter-subject factor ANNOYED revealed higher PEP values for the high-annoyed subgroup at CALM (+3.2%, P=0.022). A trend to higher PEP values also was found for ROAD (+2.8%, P=0.063) but not for RAIL. As the interaction TRACK × SEX is not of relevance, it was not considered further.

The ANOVA for HF revealed main effect on SOUND (F=21.61, P < 0.001) and demonstrated interaction ANNOYED × SEX (F=4.60, P=0.038). The first-order interactions with SOUND are: SOUND × ANNOYED (F=1.64, P=0.200), SOUND × TRACK (F=0.89, P=0.476), and SOUND × SEX (F=2.41, P=0.096). Post-hoc paired comparisons on factor SOUND revealed a decrease in HF values for RAIL (−3.4%, P < 0.001) and for ROAD (−2.9%, P < 0.001) in relation to CALM. Again no significant difference was found between RAIL and ROAD. Interaction ANNOYED × SEX was evaluated on data pooled over stress conditions. Here post-hoc comparisons at subgroup “men” revealed a trend to increased HF values (+18.7%, P=0.079) in the high-annoyed subgroup.

Salivary stress biomarker sAA

Saliva sampling was done during the first resting period and at the end of each active period, for results see [Figure 3]. In comparison to REST (103 ± 62 U/l) sAA was higher for RAIL (245 ± 167 U/l, P < 0.001), ROAD (232 ± 159 U/l, P < 0.001), and CALM (138 ± 119 U/l, P=0.006). The ANOVA revealed main effect on SOUND (F=35.95, P < 0.001) and the first-order interactions with SOUND are: SOUND × ANNOYED (F=3.72, P=0.028), SOUND × TRACK (F=1.11, P=0.358), and SOUND × SEX (F=0.57, P=0.569). Post-hoc paired comparisons on factor SOUND revealed an increase in sAA for RAIL (+78%, P < 0.001) and for ROAD (+69%, P < 0.001) in relation to CALM. No significant difference was found between RAIL and ROAD. Furthermore, post-hoc comparisons were conducted with factor ANNOYED, separately for each SOUND level. Here no significant effects were achieved.

Figure 3

Salivary stress biomarker alpha amylase (sAA) at REST and after RAIL, ROAD, and CALM; mean values with standard error

DISCUSSION

The main objective of this study was to examine extraaural effects as induced by the two traffic noise stressors (RAIL, ROAD) with same LEQ (75 dBA). Here the results show that both stressors induced congruent changes in four of the investigated variables (HR, PEP, HF, sAA), whereas in the other two variables (sBP, TPR), rather minor changes occurred. Finally, no differences between stressors RAIL and ROAD were found for all variables. These results were obtained during performing an elementary mental task. The secondary objective was to elucidate whether the disposition to traffic noise becomes reflected into the investigated variables. Here, the results showed that this was the case for variables PEP and TPR, mainly at test condition CALM.

In comparison to CALM, both noise stressors induced a chronotropic effect (accelerated HR) in parallel with an increased sympathetic drive (decrease of PEP), and a withdrawal of parasympathetic drive (decrease of HF). The changes occurred almost simultaneously with onset of a noise stressor, and after that the amplitude of the three variables remained stable over exposure time [Figure 2]. Such a fast response behavior indicates neuronal-mediated receptor activation via the two branches of the ANS projecting to the sinoatrial node. Furthermore, at the end of noise exposure, we found a relative strong activation of the SAM system, as measured by the increase in the salivary stress biomarker sAA. It therefore appears conceivable that in a later phase also humoral-mediated receptor activation contributes to increased sympathetic drive.

Several other studies reported a noise-induced chronotropic effect in combination with changed ANS parameters. In a study with white noise at 85 dBA,[18] accelerated HR was accompanied by a decrease in the HF component, corresponding to a withdrawal of parasympathetic drive. In a field study with traffic noise during sleep,[8] a significant withdrawal of parasympathetic tone was found, whereas sympathetic activity as measured by PEP was less effected. On the contrary, in the study by Lee et al.[19] with short-term white noise (5 min at levels of 50, 60, 70, and 80 dBA), there was no effect on HR and BP, but HF responded with a decrease.

In the current study, sBP slightly elevated with exposure time [Figure 2]. This effect became significant for stressor RAIL but not for stressor ROAD. Basically, an analogous increase in vascular tone or afterload is expected as causing factor; however, the TPR values did not mirror such a trend. TPR, depending on mean arterial pressure, almost remained constant during the noise exposures. Therefore, other currently unknown factors have to be responsible for the elevation in sBP. Beyond that sBP and TPR were higher in the male subgroup. However, higher BPs are quite normal for men with an age of up to 50 years.[20]

Other studies offer a diverse picture on BP. In studies with intensities above 75 dBA, basically an elevation of sBP is observed.[21] But also at intensities below 75 dBA, some elevation has been observed following hours of noise exposure during a field study.[22] In a comparable study with traffic noise at 75 dBA,[2] sBP tended to decrease against the end of exposure time. Such an effect we sometimes observed in the pretests, before implementation of the short-term memory test. Obviously vigilance becomes lowered at laboratory situations; however, under distracting conditions (CALM), such an effect is avoided. The salivary biomarker sAA shows that CALM is a relative mild stressor (+34% in relation to REST) compared with RAIL (+78% in relation to CALM) and ROAD (+69% in relation to CALM), see Figure 3.

Opposed to our hypothesis, no differences in the six variables could be determined. Although RAIL traffic noise reaches higher peak sound pressure levels, this has not led to higher stress levels as indicated by the three autonomic variables PEP, HF, and sAA. Significant interaction between SOUND and TRACK was only observed for HR. HR always was highest at the first presentation within a sound track (RAIL or ROAD), indicating a time-order effect. At least, also intersubject factors SEX and ANNOYED showed insignificant during both noise stressors. All together, these results support evidence that measures of LEQ are adequate to predict stress reactions induced by rail and road traffic noise. However, it has to be noted here that these results were obtained in the laboratory with sound samples at moderate loudness (75 dBA) at exposure times of 20 min.

Noticeable are also the effects of annoyance on variables PEP and TPR. The effects were significant during CALM and showed trends during RAIL and ROAD. Specifically, PEP was increased in the high-annoyed subgroup indicating a reduced sympathetic drive in comparison to the low-annoyed subgroup. On contrary, TPR increased in the high-annoyed subgroup indicating a higher vascular tone. At a first view, this result appears ill posed as for persons with aversions to traffic noise a higher sympathetic drive is expected. But from a physiological point of view, it makes sense that an increased vascular tone is accompanied by a reduced sympathetic drive to the heart. Further studies are necessary to reveal the genesis and relevance of this psychophysiological relationship.

Last, from a health political perspective, it is of interest to see whether there are differences in nonauditory physiological responses as induced by rail and road traffic noise with same LEQ. In some European countries, there exists a consensus that rail noise is less annoying than road noise and therefore a “rail-bonus” was defined that allows operators of railed vehicles to subtract some decibels from the measured sound emissions.[23] The current study, however, demonstrated comparable extraaural effects, which is in line with the findings of an earlier study.[24] These findings should be incorporated in forthcoming guidelines for traffic noise emissions and the protection from rail and road traffic noise.

Financial support and sponsorship

This study was financed by the Hygiene Fund of the Medical University of Graz.

Conflicts of interest

There are no conflicts of interest.