Noise, defined as unwanted sound, is a ubiquitous environmental and occupational stressor. While noise is quite common, it is a complex exposure due to its varying subjective (annoyance and sensitivity) and objective (loudness, frequency/pitch) characteristics. The effects of noise on hearing are well elucidated. However, in recent years particular interest has been in parsing out its effects on cardiovascular health. Associations of long-term noise exposures with actual disease manifestation such as hypertension, myocardial infarction, and ischemic heart disease have been observed (Basner et al., 2014; Babisch, 2011). What is less clear, is an understanding of the underlying mechanisms responsible for these cardiovascular effects as well as the role the subjective and objective components play in mediating these effects.

It is hypothesized that noise affects cardiovascular health through a stress mechanism via the autonomic nervous system and endocrine system. Over short time periods, noise exposed individuals experience increases in blood pressure, changes in heart rate variability (HRV), and the secretion of stress hormones including cortisol and amylase (Basner et al., 2014; Chang et al., 2009). Over longer time periods, continued exposure-response stress loops begin to affect the homeostasis of the human organism, giving rise to risk factors such as increased blood pressure, increased blood lipid concentrations, lower blood viscosity, and increased blood glucose concentrations that are well known for promoting the development of poor cardiovascular health. (Basner et al., 2014).

In experimental studies, the stress mechanism hypothesis has been tested using short-term exposure to noise and biological stress response measures of blood pressure, HRV, salivary amylase, and cortisol. Results of such studies are mixed. For blood pressure, while Lee et al. (2010) found no changes across noise exposed groups, Lusk et al. (2004) and Zamanian et al. (2013) observed increased blood pressure with noise exposure. Likewise, Bjor et al. (2007) found no changes in HRV between noise exposure groups; however, both Lee et al. (2010) and Kraus et al. (2013) found significant changes in HRV in groups exposed to higher noise intensities. Wagner et al. (2010) showed significant increases in both salivary cortisol and amylase after exposure to noise.

One of the major limitations of mechanistic studies of noise and stress is their implicit assumption that the sound pressure level is the most relevant characteristic. What is less known, however, is the influence of noise frequency on the stress response. The importance of considering noise frequency comes primarily from laboratory studies. These types of studies suggest that our body's organs respond to different frequencies differentially, with low-frequency noise being especially deleterious (Alves-Pereira and Branco, 2007). Studies of the effects of noise frequency on human health beyond the laboratory are quite limited. In an observational study of a population of oil mill workers in India, Kumer et al. (2008) found that with workers exposed to noise frequencies ranging from 350-700 Hz, roughly a third of them had cardiovascular problems (Kumer et al., 2008). Experimentally, a panel study found increased salivary cortisol levels in individuals exposed to low-frequency noise as compared to white noise (Waye et al., 2003).

We conducted a pilot study to investigate: (1) whether noise exposure produced acute changes in stress and cardiovascular responses; and (2) whether these responses differed based on noise frequency. Using a panel study design where participants were monitored on multiple visits during no noise, low- or high-frequency noise exposure scenarios, we evaluated changes of cardiac autonomic function as measured by HRV and blood pressure and on the endocrine system as measured by salivary cortisol and amylase.

Overall, we had a total of 13 study participants. However, three participants were excluded in our analysis—one for of violating study protocol and two because they only completed a “no noise” noise exposure. The mean age was 26 years, with a standard deviation of approximately 8 years. Of the 10 participants, 9 participated in the “no noise” exposure scenario, 7 in at least one of the low-frequency scenarios, and 9 in at least one of the high-frequency scenarios. Six participants completed all six sessions and 7 completed at least one of each of the exposure scenarios for a total of 41 person-days of monitoring. We obtained a total of 658, 181, and 81, HRV, saliva, and blood pressure measurements, respectively.

Descriptive statistics of these biological measurements throughout the course of the experiment are presented in Table 2. Mean HRV and salivary stress markers generally decreased during noise exposure, followed by an increase afterward. Pulse and systolic blood pressure increased after noise exposure, while diastolic blood pressure decreased.

The results from linear mixed model 1, which shows the association between the biological measures and exposure period, are shown in Table 3. While all HRV and salivary measures decreased during noise exposure, statistically significant declines were observed with LF and SDNN. All HRV parameters and salivary amylase had statistically significant increases after noise exposure, while cortisol continued to decrease after noise exposure. Systolic blood pressure decreased after noise exposure while pulse pressure increased. Both of these observed effects were statistically significant.

In Model 2 (Table 3), we considered the separate effects of low- and high-frequency noise exposures, using an interaction term . Similar to Model 1 results, all HRV parameters decreased during noise exposure. During low-frequency noise exposure, HF, LF, and SDNN all statistically significantly declined. During high-frequency noise exposure, only LF experienced statistically significant declines. While RMSSD decreased during both high- and low-frequency noise exposure scenarios, this effect was not statistically significant. Following noise exposure, HF, LF, RMSSD, and SDNN all increased. However, statistically significant increases were observed only during high-frequency noise exposure and only for HF.

For salivary amylase, declines were observed during low-frequency noise exposure while increases were observed during high-frequency noise exposure. However, neither of these effects were statistically significant. In contrast, there were small increases in cortisol during both low- and high-frequency noise exposures. After the noise exposure ended, there were further non statistically significant changes in levels of both amylase and cortisol.

Finally, for blood pressure, after both low- and high-frequency noise exposures, diastolic blood pressure increased while the systolic blood pressure appeared less affected and the pulse rate decreased. None of these effects were statistically significant.

The main purpose for this study was to test whether noise exposure produced acute changes in stress and cardiovascular responses and whether these responses differed based on noise frequency operating under the hypothesis that noise acts on the body as physiological stressor. This stressor disrupts homeostasis by dysregulating the hypothalamic-pituitary-adrenocortical (HPA) and sympathetic-adrenal-medullar (SAM) axis by prolonging typical “fight or flight” responses such as increases in blood pressure or the secretion of cortisol (Recio et al., 2016). For cardiovascular activity, as measured by HRV, acute noise exposure may serve to dysregulate how the SNS and PNS work in tandem to generate a suitable stress response and where lowered HRV can be viewed as the inability to manage noise induced stress adequately (Recio et al., 2016) In mode1 1 where we considered associations between biological measures and exposure period, we observed statistically significant declines in HRV parameters LF and SDNN during noise exposure. Contrary to our biological hypothesis, we observed declines in both salivary amylase and cortisol during noise exposure. However, neither of these declines were statistically significant. A statistically significant decline was observed with diastolic blood pressure and statistically significant increases were observed with pulse. When we took into account noise frequency (model 2), we gained further statistically significant declines in the HRV parameter HF during low-frequency noise exposure and lost statistically significant declines in SDNN during high-frequency noise exposure. While we did see a suggestion of recuperation of HRV after noise exposure, gains were less during low-frequency noise exposure than with high-frequency noise exposure. Therefore, reductions in HRV—which we observed during both noise exposure overall and low frequency noise exposure, specifically—suggests that such noise may compromise the ability for SNS and PNS to generate a regulated response. In particular, these HRV results suggest three important findings: (1) Low-frequency noise exposure has a more negative impact on the cardiovascular response than high-frequency noise exposure; (2) The HRV parameter LF is negatively impacted by noise exposure regardless of the frequency content of the noise exposure; (3) HRV effects resulting from low-frequency noise exposure tend to persist over time. No statistically significant changes were observed in saliva and blood pressure metrics with either low- or high-frequency noise exposure. Our results showing a cardiovascular autonomic response to noise exposure are consistent with other studies—especially when a noise exposure frequency profile was directly or indirectly considered (Sim et al., 2015; Kraus et al. 2013; Roque et al. 2013; Bjor et al. 2007; Yanagihashi et al. (1997). Sim et al. (2015) examined the effects of traffic, speech, or mixed (traffic and speech) noise (45 dB) on HRV and blood pressure in college aged men and found reductions in the HRV parameter LF, but only in the speech noise exposure group and only after noise exposure occurred. In both our pooled and separate analyses, we observed increases, albeit non-statistically significant, in LF after noise exposure.

Studies able to capture HRV activity during noise exposure tend to be consistent with our findings. For concurrent noise exposures, Kraus et al. (2013) examined the relationship between HRV in male and female adults and noise levels as study participants went about their daily routine activities. For sound levels > 65 dB(A)--the range of noise exposure intensity used in our study-- negative associations between SDNN and concurrent 5dB(A) increases in noise exposure were observed. For sound levels <65 dB(A) LF and HF HRV were negatively associated with concurrent 5 dB increases in noise. SDNN was also negatively associated but with noise lagged by 5-15 minutes. Roque et al. (2013) considered the effects of baroque and heavy metal music on HRV in healthy females. Similar to our findings, statistically significant reductions were observed with LF during exposure to heavy metal music. Bjor et al. (2007) exposed male and female subjects to four exposure conditions: only noise (broadband noise at 85 dB(A), both noise and vibration, only vibration, and a control condition. While they did not observe a cardiovascular autonomic response with noise exposure alone, decreases in LF and HF were observed during exposure to the combination of noise and vibration groups and to vibration only. Yanagihashi et al. (1997) studied HRV in young females during three different noise exposure scenarios: (1) music from a synthesizer; (2) bird sounds; (3) mechanical sounds. They observed that HF was significantly lower during mechanical sounds exposures.

Contrary to our results, Lee et al. (2010), after exposing healthy adults to broadband white noise of intensities ranging from 50 to 80 dB(A), found statistically significant increases in LF during exposure to noise at sound levels of 50, 60, 70, and 80 dB(A). In a study similar to that of Roque et al. (2013), Ferreira et al. (2015) considered the effects of classical baroque (64 – 84 dB) and heavy metal (75-84 dB) music on HRV in healthy females. SDNN and LF were found to increase after exposure to classical baroque music. With heavy metal music, after noise exposure there were increases in SDNN, RMSSD, and LF.

Similar to our findings of non-statistically significant increases in blood pressure after exposure to low-frequency noise, Sim et al. (2015) found non-statistically significant increases in both systolic and diastolic blood pressure after noise exposure. Pauvonic et al.(2014) measured blood pressure during noise exposure and found increases during noise exposure for both men and women with returns to pre-exposure levels after the noise exposure ended. Both Chang et al. (2009) and Lusk et al. (2004) simultaneously measured systolic and diastolic blood pressure and noise levels as participants went about their daily activities during their work and non-work day. Chang et al.(2009), observed that a 5 dB(A) increase in the 24 hour average environmental noise exposure significantly increased SBP and DPB in both young men and women. Lusk et al. (2004) found statistically significant associations between noise exposure and systolic and diastolic blood pressure.

For salivary amylase and cortisol, Wagner et al. (2010) saw increases in both measures after healthy male and female study participants were exposed to recorded road and rail traffic noise at a sound level of 75 dB(A). In our separate analysis we did see suggestions of increases in cortisol during noise exposure to both high- and low-frequency noise exposure, but the increases were not statistically significant.

Contrasting results between our study and others can most likely be attributed, at least in part, to differences in study design—in particular, differences in how the noise exposure was defined and differences in the time periods in which the cardiovascular and stress measures were analyzed. Noise exposures in all studies were different in terms of intensity, frequency profile, duration and context. Kraus et al.(2013), Lusk et al.(2004), and Chang et al. (2009), all looked at noise exposures as study participants went about their daily activities over a 24 hour period. Paunovic et al. (2014)., Roque et al. (2013), Ferreira et al. (2015), Yaganashi et al. (1997), all exposed their study participants to pre-recorded contextual noises in a carefully controlled experimental setting for less than an hour. Our study, along with Lee et al. (2010) and Bjor et al. (2007) exposed study participants to non-contextual white noise of varying frequencies for just under an hour. All of the experimental studies exposed participants to each noise scenario in one setting whereas we allowed at least one day of wash-out between scenarios. It can be argued that each of these differences activate sympathetic and parasympathetic activity differentially, thus producing differential cardiovascular and stress responses.

Further, differences in when cardiovascular and stress measures were analyzed may also contribute to differences in observed responses to noise exposure. We had three distinct exposure time periods—no noise, during noise, and after noise. Our results were most consistent with those who examined responses during similar time frames.

Study population differences may also explain contrasting results. Our study only included healthy, young males. The cardiovascular response to noise may differ by gender. While no gender effect was observed by Bjor et al. (2007), other studies have shown that females have a more pronounced cardiovascular response to noise (Paunovic et al. (2014), Chang et al.(2009), and Lusk et al. (2004)). Likewise, age appears to be an effect modifier of the cardiovascular response to noise with individuals ≥ 65 years old experiencing larger cardiovascular response to noise exposures with sound levels ≤65 dB(A) compared to younger individuals Kraus et al. (2013). Finally, Chang et al. (2009) found that hypertensive adults were more susceptible to noise exposure with a greater effect on systolic blood pressure.

Some limitations of our findings should be considered. First, the number of study participants was small and while we observed some statistically significant cardiovascular responses to noise, such a small sample size may have limited our ability to detect all cardiovascular responses to noise. While we considered healthy males, we did not collect information on existing stress and sensitivity to noise. Previous research suggests that sensitivity plays a significant role in how individuals biologically respond to noise exposure (Recio et al., 2016, Shepherd et al, 2010) Waye et al. (2002), for example, found that with low-frequency noise exposure, subjects who were classified as highly noise sensitive maintained higher cortisol levels compared to low-sensitive subjects, suggesting differential stress responses based on an individual's noise sensitivity. Lusk et al. (2004) found that stress level positively influenced heart rate. Additionally, we used a comparatively narrow-band, non-contextual noise source, which may not have been perceived by our study participants as noise, but rather sound. The difference in perception may lead to differences in response. Our noise exposure only varied by frequency profile, which does not allow us to consider differences in response due to different intensities of noise.

From our results, we know that there is a statistically significant decline in HRV with low-frequency noise exposure. How this relates to clinical outcomes is unclear. We did observe recuperation of negative dips in HRV after the noise exposure ended. Acutely, effects include typical fight or flight reactions such as increased blood pressure and cardiac output while indirect effects include hormone release (Kraus et al., 2013). Continued stimulation of these two systems leads to disease manifestation, which includes hypertension, myocardial infarction, arteriosclerosis, ischemic heart disease, and stroke (Babisch, 2014). Epidemiological studies do suggest that chronic exposure to noises with low-frequency profiles such as aircraft, rail, and road traffic noise is associated with negative cardiovascular and stress responses such as elevated cortisol (Selander et al., 2009); blood pressure (Haralabidis et al., 2009; Dravata et al., 2012); hypertension (Bluhm et al., 2007; Bodin et al., 2009; Babisch et al., 2005); myocardial infarction (Babisch et al., 2005; Selander et al., 2009); medication use (Floud et al. 2011); cardiovascular related hospital admissions (Hansell et al., 2013; Corriea et al., 2013), and mortality (Hansell et al., 2013).

And while our sample size was quite small, our study design allowed each individual to serve as his own control, thereby enabling the examination of individual differences. We were also able to focus on the effect of noise frequency, which eliminates bias that may be introduced by known environmental noise sources. The study is also one of only a few that has looked at the effects of audible noise frequencies on such a wide variety of stress and cardiovascular outcomes.

In conclusion, the study adds to our understanding of the acoustic characteristic that drives the cardiovascular autonomic response to noise exposure. The results suggest that low-frequency noise in particular negatively impacts heart rate variability and these impacts may persist after noise exposure ends. In future studies, the dominant frequencies of noise should be considered when evaluating the cardiovascular health effects of noise.