Home 
Calculation of the physiologically developed nighttime noise index from existing noise indices Correspondence Address: Background: The effects of noise on sleep and health have been evaluated in earlier studies using noise indices chosen on the basis of practical considerations and not on the physiologic mechanisms of sleep disturbance due to noise exposure. We investigated the neurophysiologic mechanisms of sleep and found that the arithmetic mean of the sound levels above a threshold ([INSIDE:1]) may be used effectively to evaluate the effects of nighttime noise exposure. However, the efficacy of using [INSIDE:2] has only been confirmed in a specific setting; therefore, the reliability of [INSIDE:3] in other situations should be investigated by applying it in epidemiologic studies. In this study, we aimed to obtain an alternative equation for calculating [INSIDE:4] from existing noise indices, given that detailed information on fluctuations in sound levels, needed to calculate [INSIDE:5] according to the definition, is not readily available. Materials and Methods: We examined statistical relationships among noise indices namely [INSIDE:6], the number of noise events above 60 and 70 dB (N_{60} and N_{70}), and the night equivalent sound level (L_{night}). The study area was around the Kadena military airfield, where the sound levels were recorded at noise monitoring stations. Results: [INSIDE:7] showed a very strong correlation with N_{60} and N_{70} but not with L_{night}. Among regression equations, an equation representing N_{60} and the interaction between N_{60} and L_{night}, which is equivalent to the product of N_{60} and a linear expression of L_{night}, showed the highest prediction capability. Conclusion: We obtained a regression equation to calculate [INSIDE:8] from N_{60} and L_{night}. Although this alternative equation for [INSIDE:9] is only applicable for military aircraft noise, it will aid the reanalysis of existing epidemiologic studies as well as further investigations on the relationship between noise exposure and health effects.
Introduction In a previous study, we developed a new nighttime noise index to evaluate the effects on sleep based on the neurophysiologic mechanisms.[1] The new index can be applied to arbitrary noise exposure and could evaluate sleep disturbance more accurately than existing noise indices. As a noise index, night equivalent sound level (Lnight) has been widely used in epidemiologic studies and strategic noise mappings;[2],[3],[4] however, the new index may improve the evaluation of nighttime noise. The effects of noise on sleep are considered to be largely dependent on the number of noise events and fluctuations in sound levels;[4],[5],[6],[7] however, they have rarely been evaluated in existing studies. The Lnight has been selected for practical reasons, whereas a recent review also indicates that there are inconsistencies in the exposure–response relationships between Lnight and sleep disturbance.[4] The inconsistencies may be caused by the fact that Lnight is not based on the physiologic mechanisms of the effects of noise on sleep. Supplementary noise indices to evaluate the contribution of the intermittency of noise exposure to sleep disturbance have also been proposed;[8],[9],[10],[11],[12] however, physiologic validity has not been observed in them. We thus integrated the neurophysiologic theory and epidemiologic findings and developed the new nighttime noise index.[1] The new index, which is the arithmetic mean of the sound levels above 60 dB ([INSIDE:10]), is fundamentally different from the existing indices as it is based on neurophysiology and can be applied to arbitrary noise exposure. To our knowledge, it is the only index that evaluates the effect of noise on sleep on the basis of neurophysiology. However, the efficacy of [INSIDE:11] has only been confirmed in a specific epidemiologic study. Its reliability should be further investigated through epidemiologic studies, although detailed information about fluctuations in sound level, necessary to calculate [INSIDE:12] according to the definition, is not readily available. Hence, an easy method to calculate [INSIDE:13] is required. The aim of our study was to obtain an alternative equation linking existing noise indices and the new index developed in our previous study for reanalyzing existing epidemiologic studies. Using soundlevel data recorded around an airfield, we calculated noise indices, and statistically examined the relationships among them. Materials and Methods The study aimed to obtain an alternative equation to calculate [INSIDE:14] from existing indices. The definition of [INSIDE:15] is as follows: [INLINE:1] where, t is time, L(t) (dB) is the sound level, T60dB is the time interval for which the sound level is above 60 dB, and Tnight is the duration of nighttime for which the effects of noise exposure are evaluated. L(t) is necessary to calculate [INSIDE:16] according to this definition; however, it is not readily available from existing studies. [INSIDE:17] is different from existing noise indices used for practical purposes as it is derived from neurophysiology. We investigated the relationship between the sound level and the electrical potential of the brainstem according to neurophysiology[13],[14] because sleep and wakefulness are regulated by the nuclei in the brainstem,[15] and the potential of the brainstem corresponding to wakefulness will be greater. Then, we integrated the neurophysiologic theory with the results of an epidemiologic study, eventually deriving a new noise index [INSIDE:18] to evaluate the effects of noise on sleep.[1] Existing noise indices that have been widely used in epidemiologic studies (e.g., Lnight) cannot provide an analytical relationship with [INSIDE:19]. Therefore, we numerically calculated noise indices from measurements of the sound level and statistically investigated the relationships among them. The study area was located surrounding the Kadena military airfield, Okinawa prefecture, Japan ([Figure 1]). Numerous military aircraft takeoff from and land on the two runways around the clock. The local governments have set up 21 noise monitoring stations around the airfield and the sound level was measured every 1 second with a precision soundlevel meter.{Figure 1} A previous survey indicated that the night flight paths were predominantly aligned with the centerlines of the two runways.[16] We selected 10 stations near the extended lines of the runways to investigate noise exposure from nighttime aircraft activities. At other stations, there was a higher likelihood of interference due to road traffic noise during nighttime. The soundlevel data at the 10 stations were provided by the local governments of Okinawa Prefecture, Okinawa City, Kadena Town, and Chatan Town. The study was conducted from April 1, 2014 to March 31, 2015. We defined 22:00 to 7:00 as the nighttime and noise events that exceeded 60 dB for more than 5 seconds as aircraft noise events. We calculated [INSIDE:20], Lnight, and the number of noise events above 60 and 70 dB per night (N60 and N70) at each monitoring point. We statistically investigated the relationships among the noise indices. First, we examined the correlations among them and then conducted a multivariate regression analysis, where we calculated the secondorder corrected Akaike’s Information Criteria (AICc)[17] to evaluate the prediction capability of the regression equations. Moreover, we estimated the risk of sleep disturbance that would be related to health risk, using the regression equation. According to the neurophysiologic theory, the relative risk of the effects of noise on sleep (RR) proportionally increases with the new index.[1] We employed the following equation derived from our previous study on the onset of motility during sleep due to nighttime noise exposure: [INLINE:2] All statistical procedures were conducted with 3.6.1 (Vienna: R Core Team, 2019).[18] Results [Table 1] summarizes the characteristics of noise measurements at the monitoring points around the Kadena military airfield. We excluded nights with storm warnings and warnings for strong wind, announced by the Japan Meteorologic Agency. In addition, there were several nights for which the data were not available. The table also lists maximum sound levels (Lmax), of which high values at all monitoring stations suggest that the aircraft noise may disturb sleep around the airfield. Several differences were observed in the noise indices investigated in the present study. For instance, N60 and [INSIDE:21] at P10 were higher than at P15, but Lnight displays the opposite trend. The number of noise events makes only a relatively small contribution to Lnight.{Table 1} [Figure 2] shows the correlations of the noise indices. N60 and N70 showed a high correlation with [INSIDE:22], whereas Lnight showed a relatively weak correlation.{Figure 2} Although the univariate regressions provide significant insights into the relationships among noise indices, the evaluation of the prediction capability of the multivariate regression equations is the most important for calculating [INSIDE:23]. [Table 2] summarizes the fitting results of the multivariate regression analysis. As the correlation between N60 and N70 was high, we excluded N70 from the regression analysis. The regression equation that includes N60 and the interaction between N60 and Lnight minimized the AICc.{Table 2} Thus, we obtained the following regression equation for [INSIDE:24] which has the highest prediction capability: [INLINE:3] Notably, a threshold (43.2 dB in Lnight) identified in the equation could contain significant errors as it is the extrapolation of the data, of which sound levels were relatively high and sample size was small. Recent guidelines indicate that noise from commercial aircraft may cause sleep disturbance even at 40 dB in Lnight.[2],[4] The coefficients of the equation should be revisited in the case where different noise sources such as commercial aircraft are investigated. [Figure 3] shows the close correspondence between predicted values using Eq. (3), and observed values of [INSIDE:25].{Figure 3} [Figure 4] shows the estimated risk of sleep disturbance (onset of motility) calculated from the multivariate regression equation for [INSIDE:26] (Eq. (3)), and the exposure–response relationship between [INSIDE:27] and relative risk of the sleep effects of noise (Eq. (2)). The risk proportionally increased with Lnight and N60.{Figure 4} Discussion For evaluating the health effects of noise, the effects of noise on sleep should be investigated. The newly developed nighttime noise index based on neurophysiology namely [INSIDE:28] can be a viable option to evaluate it. However, the efficacy of [INSIDE:29] remains unclear. In this study, we investigated statistical relationships between the new index and existing indices, for employing the new index in existing epidemiologic studies and for elucidating the relationship between noise exposure and its effects on sleep and health. [INSIDE:30] showed a strong correlation with the number of noise events (N60 and N70) but not with the mean acoustic intensity (Lnight). This indicates that N60 and N70 are good estimators for [INSIDE:31] and that the energybased evaluation of noise exposure is not a good approach to calculate [INSIDE:32]. As shown in existing studies,[4],[5],[6],[7] the number of noise events would be very important for evaluating the effects of noise on sleep. Further investigations are required for the difference between [INSIDE:33] and Lnight; however, providing that the neurophysiologic effects of noise on sleep can be indexed by [INSIDE:34], a better explanation regarding the effects of noise on sleep might be provided by [INSIDE:35] than Lnight owing to an inconsistent relationship with sleep disturbances and Lnight.[4] Multivariate regression analysis showed that a regression equation represented by the product of N60 and a linear expression of Lnight exhibited the minimum AICc value and the highest prediction capability. The significance of this study lies in accurately estimating [INSIDE:36] from the widely used indices N60 and Lnight.[19],[20] As several studies have introduced an additional noise index to Lnight for evaluating the intermittency of noise exposure,[8],[9],[10],[11],[12] the results of this study suggest that the effects of noise on sleep could be more accurately evaluated with the combination of N60 and Lnight. Applying [INSIDE:37] to epidemiologic studies can contribute toward proving its reliability, as well as elucidating the relationship between noise exposure and its effects on sleep and health. The regression equation eventually enabled estimation of the risk of sleep disturbance using N60 and Lnight, owing to the proportional relationship between [INSIDE:38] and the risk. It is worth noting that the risk of the onset of motility may vary even if N60 or Lnight is the same. That is, these single noise indices may fail to evaluate the effects of noise on sleep. The estimations of sleep disturbance risk are based on our investigation of the onset of motility, and the magnitude of the increase in the risk may contain a significant error; however, as the effects of noise on sleep could be linked with the adverse health effects due to noise exposure,[21] the estimation of the risk of sleep disturbance may help in evaluating the health risks due to noise exposure. The main limitation of the present study is that the regression analysis is based only on measurements around the military airfield. The characteristics of the fluctuations in the sound level of military aircraft would be different from those of commercial aircraft. Moreover, we did not consider other noise sources such as road traffic and railway. The mean sound level and the number of noise events may play an important role; however, further investigations are necessary to determine their relationships. Moreover, [INSIDE:39] could underestimate the effects of noise on sleep because it assumes the threshold of the soundlevel outdoor to be 60 dB.[1] Determining this threshold remains a challenge. Conclusions In this study, to obtain an alternative equation for calculating the newly developed nighttime noise index based on neurophysiology, [INSIDE:40], we statistically investigated the relationship between the new index and existing noise indices using the measurements of sound levels around the Kadena military airfield. The product of N60 and a linear expression of Lnight showed high accuracy in predicting [INSIDE:41]. Reanalysis of existing epidemiologic studies using the regression equation would be useful for investigating the reliability of the new index and elucidating the exposure–response relationship between noise exposure and health effects on the basis of neurophysiology. Acknowledgements We would like thank to Okinawa Prefecture, Okinawa City, Kadena Town, and Chatan Town for providing the data of aircraft noise. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest. References


