Negative impact of noise and noise sensitivity on mental health in childhood.

INTRODUCTION: Noise and noise sensitivity have negative effects on mental health and are not well-studied in children and adolescents. In this study, we investigated these effects in the aforementioned population with respect to sociodemographic variables and environmental factors.
MATERIALS AND METHODS: In this population-based study conducted in two large cities in South Korea, 918 elementary and middle-school students were included. After direct measurements at the selected sites, a noise map was created using an interpolation method. The road traffic noise of the participants' residential areas was calculated based on this noise map. Noise sensitivity was assessed on an 11-point Likert scale. Using multivariate logistic regression, we investigated the relationship among noise, noise sensitivity, and the Child Behavior Checklist. Further analyses were performed subdividing the data according to household income levels.
RESULTS: Noise sensitivity was significantly associated with internalizing, externalizing, and total behavioral problems. Noise was positively associated with total behavioral problems. In the low-income group, the degree of association with problem behaviors was higher, whereas the relationship between noise sensitivity and externalization problems disappeared in the high-income group.
CONCLUSION: Noise and noise sensitivity are negatively associated with the mental health of children and adolescents, particularly in low-income groups. The findings of this study suggest that noise sensitivity and socioeconomic status should be considered in coping with negative effects of noise in children and adolescents.

Introduction

Noise negatively affects auditory and nonauditory health in adults.[12] With respect to the nonauditory health effects of noise, an association among noise exposure and hypertension, cardiovascular diseases, and stroke has been reported.[23456] Although research on the effect of noise on mental health has yielded inconclusive results,[78] studies have reported that noise exposure is associated with emotional distress,[2] sleep disturbances,[91011] psychosomatic disorders,[1213] and psychiatric hospital admission rates.[1415] Because of these negative effects, noise can impair the quality of life.[1617] Among various noise sources, road traffic noise is of special interest, considering its generally wide and usually long exposure.[18] Traffic noise was cited as the second most influential environmental risk factor after particles in a recent European study.[19]

Noise sensitivity has firstly been defined as a “general measure of attitudes toward noise” by Anderson.[20] Thereafter, different researchers have used different definitions of noise sensitivity; however, it is generally accepted that noise sensitivity refers to a stable trait or internal state of an individual.[212223] Although some researchers insist that noise sensitivity reflects subjective sensitivities toward the environment in general, rather than toward noise alone[24]; it is more likely that noise sensitivity reflects a specific discriminating sensitivity toward noise than toward the general environment.[2526]

It has been suggested that noise sensitivity rather than the actual noise level is the most important factor in noise-related health effects. In a previous study, noise sensitivity accounted for 10% to 26% of noise-induced annoyance.[27] Furthermore, it was reported that the nonauditory health effects of noise manifested only in a highly noise-sensitive group.[28] In individuals with high noise sensitivity, hypertension and emphysema are more frequent, whereas cardiovascular mortality levels are increased in noise-sensitive women. Studies have reported an association between noise sensitivity and various mental health-related factors, such as anxiety, depression, higher benzodiazepine usage, and future psychiatric disorder.[29303132]

Studies in children and adolescents have also report that environmental noise has a negative impact on children’s health.[3334] In such studies, children living in noisy neighborhoods complained of more stress symptoms than did those living in silent environments.[33] Moreover, children who considered their classroom as noisier were found to have higher diastolic blood pressure in the Los Angeles Airport Study.[34] However, studies have also reported no association between noise and children’s health. For example, Haines et al.[3536] found no association between airplane noise and self-reported health status, such as headache and tiredness, and Stansfeld et al.[37] reported that neither airplane noise nor road-traffic noise are associated with children’s self-reported health status.

Studies investigating the effect of noise sensitivity in children are insufficient. Only one study reported on noise annoyance in children, not noise sensitivity, showing that a group, which was annoyed by airplane noise had more self-reported health symptoms and negative results on neurobehavioral assessments than a group that was not annoyed by noise.[38]

Children and adolescents are less able to cope with stress compared with adults.[37] It is therefore assumed that children are more vulnerable to environmental stresses, such as noise, than are adults. In addition, childhood mental health problems can have long-term negative implications by affecting individuals both academically and occupationally.[3940] It is thus important to examine how noise-related variables affect children’s and adolescents’ mental health and to screen for children and adolescents who are vulnerable. However, research on the relationship between noise and the mental health of children and adolescents, especially regarding noise sensitivity, is currently insufficient. In the present population-based study, we investigated the effects of noise and noise sensitivity in children and adolescents on their mental health as evaluated by the Child Behavior Checklist (CBCL). We hypothesized that high levels of both noise and noise sensitivity are negatively associated with the mental health of children and adolescents, and that the effects of these noise-related variables depend on socioenvironmental factors.

Subjects and Methods

Study population

The present study was conducted between June and August 2016. We selected four elementary schools and four middle schools in Yangcheon-gu, Seoul, and in Nam-gu, Ulsan, and 120 students and their parents at each school (n = 960) were enrolled. When choosing schools in each region, we selected the school with the lowest noise level and the school with the highest noise level based on the noise level in the noise map.

A questionnaire was distributed to students and parents in each school and was sent back to the school after completion. All participants agreed to participate and provided informed consent. A total of 918 of the 960 participants, excluding 42 participants with incomplete survey data, were finally included in this study. This study was approved by the Institutional Review Board of Ulsan University Hospital (2014-08-008).

Demographic characteristics

We investigated basic demographic characteristic, such as age, sex, height, weight, socioeconomic factors, that may affect the children’s mental health and noise exposure and factors that may affect their development. As socioeconomic variables, parental educational level, monthly household income, residential environment, smartphone usage time, and computer gaming time were included. The average monthly income was categorized as <4 million Korean won (approximately 4000 US dollars) or ≥4 million won, based on the average income of households with two or more members in cities in South Korea, which amounted to 4.3 million won in 2015. As development-related variables, maternal illness during pregnancy (diabetes, hypertension, preeclampsia, and thyroid disease), parental age at birth, premature birth, low birth weight at birth, breastfeeding, parental smoking status, and passive smoking were included. The psychiatric history was obtained by asking parents to indicate the child’s illness on the questionnaire.

Noise sensitivity

Noise sensitivity was assessed using a visual analog scale that had been translated according to the International Organization for Standardization Technical Specification 15666 (2003). Noise sensitivity was assessed by the parents of each participant, using single-item questionnaires. On an 11-point Likert scale, scores of 0 and 10 points indicated the lowest and highest sensitivity, respectively. Noise sensitivity is determined after discussions between parents and child. Parents ask the child about the degree of noise sensitivity, and parents make the final decision on the score.

Noise level

The indicator of the noise level in the present study was the day–night average sound level (Ldn) of road-traffic noise. Ldn is an average of 24-h noise that applies a 10-dB penalty to noise at nighttime (10 pm–7 am). It has been used by US government agencies since the 1970s as an indicator for the assessment of the impact of environmental noise.[41] In this study, the level of road traffic noise at the exterior wall of a residential building was calculated using noise prediction software (CadnaA; DataKustik, Gilching, Germany) based on a noise map. The noise map was created in 2014 by a research team that modeled the terrain, buildings, and roads of the study area, measured the traffic volume, the large-car ratio, and speed limits, and verified the difference between the predicted and measured values.

Child Behavior Checklist

The CBCL 6 to 18 has been developed to identify behavioral problems in children via their parents’ observation. It has been translated and normalized in many Eastern and Western cultures, and its reliability and validity as an effective screening tool is well established.[42] It consists of 120 items that are assessed on a 3-point scale, from 0 to 2, judging whether the child or adolescent showed the described behavior in the past 6 months. The problem behavior scale is the most critical part of the CBCL 6 to 18 and includes eight syndrome scales (Anxious/Depressed, Withdrawn/Depressed, Somatic Complaints, Social Problems, Thought Problems, Attention Problems, Rule-Breaking Behavior, Aggressive Behavior) as well as broadband scales consisting of the sum of these subscales (internalizing, externalizing, total problem score). The internalizing problems scale refers to behaviors that are overly controlled, such as passive and restricted behaviors, and consist of the sum of the Anxious/Depressed, Withdrawn/Depressed, and Somatic Complaints subscales. The externalizing problem scale refers to undercontrolled behavior and consists of the sum of the Rule-Breaking Behavior and Aggressive Behavior subscales. The total problems score is the sum of the scores of all problem items and represents the extent of the overall behavioral problems. The raw score is converted into a T score using the mean, standard deviation, and percentile distribution of the raw score. A T score of 64 or higher is classified as the clinical range for internalizing, externalizing, and total behavioral problems.

Statistical analysis

The demographic characteristics, socioeconomic variables, and variables that may affect the development of the participants in the high- and low-noise sensitivity groups were compared according to the median value (noise sensitivity = 3) of the study population. A t test was used for continuous variables, the Chi-squared or Fisher’s exact test was used for categorical variables, and the Mann–Whitney U test was used for ordinal variables.

Logistic regression analysis was performed on the internalizing, externalizing, and total behavioral problems scales, which are the broadband scales of the CBCL 6 to 18. Before multivariate analysis, a univariate analysis was used to examine the association of variables with behavioral problems and to select covariates for the multivariate analysis. Variables that not only showed moderate associations (P < 0.200) with behavioral problems in the univariate analysis were selected as covariates for multivariate logistic regression. The adjusted models included the variables age, sex, monthly income, premature birth, maternal age at birth, passive smoking, disease during pregnancy (hypertension and preeclampsia), and mental disorders [attention deficit hyperactivity disorder, attention deficit hyperactivity disorder (ADHD), tic disorder, and conduct disorder]. The results are presented as odds ratios (ORs) and 95% confidence intervals (CIs). In the regression analysis, an evaluation of the variance inflation factors of the independent variables (Ldn = 1.05, age = 1.03, sex = 1.03, income = 1.07, premature birth = 1.07, maternal age at birth = 1.07, passive smoking = 1.02, hypertension during pregnancy = 1.08, preeclampsia = 1.11, ADHD = 1.50, tic disorder = 1.48, and conduct disorder = 1.64) confirmed that multicollinearity was not an issue.

To investigate the relationship between behavioral problems and noise exposure in schools, the schools in each city were classified as high and low noise level schools, and the proportion of behavioral problems in each group was compared using a Chi-squared test.

To examine the effect of income on the association between noise sensitivity and behavioral problems, we evaluate the presence of interaction between noise sensitivity and household income via inclusion of noise sensitivity by income interaction terms in the multivariable models and performed additional analyses where we stratified the study population into low- and high-income groups. In addition, the differences in socioeconomic as well as birth- and developmental-related variables were compared between the low and high groups. A t test was used for continuous variables, and the Chi-squared or Fisher’s exact test was used for categorical variables; the Mann–Whitney U test was used for ordinal variables.

For the sensitivity analysis, all main analyses were performed as described while excluding those participants who had lived in their respective residential area for less than 1 year. For all tests, statistical significance was set at P < 0.05 (two-tailed). Data were analyzed using IBM SPSS Statistics for Windows, Version 23.0 (IBM; SPSS Inc., Chicago, Illinois, USA).

Results

The participants had a mean age of 11.47 ± 1.54 years (age range, 9–14 years): 51.6% were in the elementary school, 48.4% were in the middle school, and 46.5% were boys. The proportion of students in the high-income group was 75.6%. The mean noise level (Ldn) of all schools was 56.84 ± 9.34 dB, and the mean noise sensitivity was 3.31 ± 2.03. There was no significant difference between the low- and high-noise groups with respect to demographic variables, birth-related variables, and medical and mental health status variables [Table 1]. The participants spent an average of 12.97 h on weekdays and 16.48 h on weekends at home. Supplement Table 2 presents the noise level at each school. There were no significant differences in behavioral problems according to the noise level at each school [Supplement Table 3].

Table 1

Demographic and socioeconomic variables of all participants according to noise sensitivity

Characteristic		Total (n = 918)	Low NS (n = 334)	High NS (n = 580)	t Test or χ 2 test
Age (years)		11.47 ± 1.54	11.47 ± 1.54	11.47 ± 1.54	0.992
Sex					0.091
	Boy	427 (46.5)	144 (42.9)	283 (48.6)
	Girl	491 (53.5)	192 (57.1)	299 (51.4)
Education level
	Elementary school	474 (51.6)	175 (52.1)	299 (51.4)	0.539
	Middle school	444 (48.4)	161 (47.9)	283 (48.6)	0.553
Monthly income (won)					0.390
	<4 million	221 (24.4)	75 (22.8)	146 (25.3)
	≥4 million	684 (75.6)	254 (77.2)	430 (74.7)
Low birth weight		36 (4.0)	15 (4.5)	21 (3.7)	0.528
Premature birth		41 (4.5)	20 (6)	21 (3.7)	0.099
Maternal age at birth (years)		30.28 ± 3.74	30.56 ± 3.34	30.12 ± 3.94	0.443
Passive smoking		127 (13.9)	45 (13.5)	82 (14.1)	0.780
Maternal illness during pregnancy
	Diabetes	15 (1.6)	7 (2.1)	8 (1.4)	0.415
	Hypertension	8 (0.9)	2 (0.6)	6 (1)	0.494
	Preeclamsia	24 (2.6)	7 (2.1)	17 (2.9)	0.448
	Thyroid disease	12 (1.3)	3 (0.9)	9 (1.5)	0.401
Mental disorders
	ADHD	15 (1.7)	3 (0.9)	12 (2.1)	0.172
	Specific learning disorder	8 (0.9)	3 (0.9)	5 (0.9)	0.973
	Communication disorder	14 (1.5)	4 (1.2)	10 (1.7)	0.515
	Intellectual disability	4 (0.4)	1 (0.3)	3 (0.5)	0.622
	Autism spectrum disorder	3 (0.3)	1 (0.3)	2 (0.3)	0.899
	Tic disorder	15 (1.7)	4 (1.2)	11 (1.9)	0.410
	Depression	4 (0.4)	1 (0.3)	3 (0.5)	0.612
	Conduct disorder	4 (0.4)	1 (0.3)	3 (0.5)	0.612
Noise level (dBA)	Ldn	56.84 ± 9.34	57.01 ± 9.05	56.74 ± 9.53	0.699
NS		3.31 ± 2.03	2.18 ± 1.31	5.67 ± 0.95	<0.001

ADHD = attention deficit hyperactivity disorder, Ldn = day–night equivalent level, NS = noise sensitivity.

Table 2

Noise level at each school

City	Grade	Name	Noise level (dB)
Seoul	Elementary school	A	49.4
		B	60.8
	Middle school	C	54.1
		D	61.9
Ulsan	Elementary school	E	50.7
		F	59.2
	Middle school	G	49.5
		H	62.0

Table 3

Comparison of behavioral problem ratio according to the noise level in schools

City	Noise level	n	Internalizing problem (%)	P value	Externalizing problem (%)	P value	Total behavior problem (%)	P value
Seoul	Low noise level (A,C)	222	3.6	0.994	3.2	0.484	1.8	0.396
	High noise level (B,D)	195	3.6		2.1		3.1
Ulsan	Low noise level (E,G)	231	4.8	0.957	4.8	0.957	3.9	0.853
	High noise level (F,H)	279	4.7		4.7		3.6

Univariate analysis

A univariate logistic regression analysis was performed using each variable as an independent variable and CBCL internalizing, externalizing, and total behavioral problems as dependent variables [Table 4]. The variables significantly associated with internalizing problems were noise sensitivity (OR = 1.35; 95% CI: 1.15, 1.60), passive smoking (OR = 3.16; 95% CI: 1.55, 6.46), preeclampsia (OR = 3.77; 95% CI: 1.07, 13.30), and ADHD (OR = 6.29; 95% CI: 1.7, 23.32). The variables significantly associated with externalizing problems were noise sensitivity (OR = 1.29; 95% CI: 1.09, 1.52), age (OR = 1.26; 95% CI: 1.00, 1.59), ADHD (OR = 10.41; 95% CI: 3.13, 34.60), autism spectrum disorder (OR = 13.15; 95% CI: 1.16, 148.72), and conduct disorder (OR = 27.13; 95% CI: 3.70, 198.73). The variables significantly associated with total behavioral problems were noise sensitivity (OR = 1.37; 95% CI: 1.13, 1.65), noise level (OR = 1.07; 95% CI: 1.01, 1.13), maternal age at birth (OR = 0.90; 95% CI: 0.81, 0.99), passive smoking (OR = 3.06; 95% CI: 1.35, 6.93), ADHD (OR = 8.64; 95% CI: 2.29, 32.54), and conduct disorder (OR = 10.78; 95% CI: 1.09, 107).

Table 4

Univariate logistic regression analysis of each behavioral problem and candidate variables

Variable			Internalizing problem		P value		Externalizing problem	P value	Total problem	P value
Noise sensitivity				1.35 (1.15–1.6)		<0.001	1.29 (1.09–1.52)	0.004	1.37 (1.13–1.65)	0.001
Noise level (Ldn)				1.01 (0.97–1.05)		0.652	1.03 (0.99–1.08)	0.138	1.07 (1.01–1.13)	0.024
Age				0.92 (0.75–1.14)		0.461	1.26 (1.00–1.59)	0.047	1.2 (0.94–1.54)	0.146
Sex				0.81 (0.43–1.54)		0.526	1.31 (0.66–2.60)	0.446	0.8 (0.38–1.68)	0.554
Monthly income				0.59 (0.29–1.17)		0.131	0.77 (0.36–1.64)	0.503	0.58 (0.26–1.26)	0.168
Low birth weight				0.69 (0.09–5.22)		0.723	NA	NA	0.95 (0.13–7.25)	0.964
Premature birth				0.61 (0.08–4.54)		0.627	2.26 (0.66–7.75)	0.195	1.78 (0.41–7.82)	0.443
Maternal age at birth				0.99 (0.9–1.09)		0.835	0.95 (0.86–1.04)	0.256	0.9 (0.81–0.99)	0.033
Passive smoking				3.16 (1.55–6.46)		0.002	1.96 (0.87–4.42)	0.107	3.06 (1.35–6.93)	0.007
Maternal illness during pregnancy
		Diabetes		1.71 (0.22–13.35)		0.61	NA	NA	2.3 (0.29–18.14)	0.429
		Hypertension		3.44 (0.41–28.74)		0.253	3.77 (0.45–31.54)	0.221	4.64 (0.55–39.05)	0.158
		Preeclmasia		3.77 (1.07–13.3)		0.039	1.18 (0.15–9)	0.875	1.45 (0.19–11.15)	0.721
Mental disorders
		ADHD		6.29 (1.7–23.32)		0.006	10.41 (3.13–34.6)	<0.001	8.64 (2.29–32.54)	0.001
		Specific learning disorder		3.41 (0.41–28.41)		0.258	3.73 (0.45–31.18)	0.225	NA	NA
		Communication disorder		NA		NA	2 (0.25–15.71)	0.512	NA	NA
		Intellectual disability		NA		NA	8.76 (0.89–86.46)	0.063	NA	NA
		Autism spectrum disorder		NA		NA	13.15 (1.16–148.72)	0.037	NA	NA
		Tic disorder		3.75 (0.82–17.28)		0.089	4.12 (0.89–19.03)	0.07	2.28 (0.29–17.98)	0.434
		Depression		NA		NA	8.76 (0.89–86.46)	0.063	NA	NA
		Conduct disorder		8 (0.81–78.81)		0.075	27.13 (3.7–198.73)	0.001	10.78 (1.09–107)	0.042

ADHD = attention deficit hyperactivity disorder, CBCL = Child Behavior Checklist, CI = confidence interval, Ldn = day–night equivalent level, NS = noise sensitivity, OR = odds ratio.

Behavioral problems

The results of the analysis of the association among noise level, noise sensitivity, related covariates, and behavioral problems are presented in Tables 5–7. Noise sensitivity was significantly associated with internalizing, externalizing, and total behavioral problems, even after adjustment for age, sex, monthly income, premature birth, maternal age at birth, passive smoking, maternal illness during pregnancy, and mental disorders in the multivariate analysis. In contrast, the noise level was significantly associated with total behavioral problems after adjustment for all covariates; however, it did not show any significant association with internalizing and externalizing problems. Monthly household income was not significantly associated with internalizing, externalizing, and total behavioral problems.

Table 5

Associations between internalizing problems of the CBCL and noise-related variables

	Model 1		Model 2		Model 3		Model 4
	OR (95% CI)	P value	aOR (95% CI)	P value	aOR (95% CI)	P value	aOR (95% CI)	P value
Noise sensitivity	1.41 (1.17–1.71)	< 0.001	1.43 (1.17–1.73)	< 0.001	1.45 (1.18–1.79)	0.001	1.42 (1.15–1.75)	0.001
Noise level (Ldn)	1.01 (0.97–1.05)	0.676	1.01 (0.97–1.06)	0.556	1.02 (0.98–1.06)	0.423	1.02 (0.97–1.06)	0.455
Age			0.86 (0.66–1.11)	0.242	0.86 (0.66–1.12)	0.268	0.87 (0.66–1.14)	0.307
Sex			1.11 (0.52–2.4)	0.786	1.1 (0.5–2.43)	0.816	1.02 (0.45–2.3)	0.971
Monthly income			0.74 (0.31–1.74)	0.49	0.95 (0.38–2.4)	0.918	1.07 (0.41–2.79)	0.89
Premature birth					NA	NA	NA	NA
Maternal age at birth					1.00 (0.90–1.12)	0.977	1.00 (0.90–1.12)	0.981
Passive smoking					2.01 (0.15–26.5)	0.597	2.26 (0.17–29.62)	0.535
Martenal illness during pregnancy
Hypertension					6.49 (1.45–29.05)	0.014	5.75 (1.22–27.12)	0.027
Preeclmasia					3.77 (1.62–8.77)	0.002	3.83 (1.64–8.92)	0.002
Mental disorders
ADHD							3.55 (0.49–25.96)	0.212
Tic disorder							1.14 (0.08–16.61)	0.922
Conduct disorder							NA	NA

ADHD = attention deficit hyperactivity disorder, aOR = adjusted OR, CBCL = Child Behavior Checklist, CI = confidence interval, Ldn = day–night equivalent level, NS = noise sensitivity, OR = odds ratio.

Table 7

Multivariate logistic regression analysis of CBCL total problems and noise-related variables

		Model 1		Model 2		Model 3		Model 4
		OR (95% CI)	P value	aOR (95% CI)	P value	aOR (95% CI)	P value	aOR (95% CI)	P value
Noise sensitivity		1.36 (1.09–1.71)	0.008	1.36 (1.08–1.71)	0.008	1.35 (1.06–1.72)	0.014	1.33 (1.04–1.70)	0.022
Noise level (Ldn)		1.07 (1.01–1.14)	0.024	1.07 (1.01–1.14)	0.024	1.07 (1.01–1.14)	0.034	1.08 (1.01–1.15)	0.034
Age				1.25 (0.93–1.69)	0.141	1.24 (0.91–1.68)	0.168	1.27 (0.93–1.75)	0.135
Sex				1.23 (0.5–3.04)	0.652	1.26 (0.5–3.18)	0.626	1.22 (0.47–3.13)	0.683
Monthly income				0.82 (0.29–2.34)	0.714	1.2 (0.38–3.82)	0.762	1.29 (0.4–4.19)	0.675
Premature birth						1.54 (0.18–13.24)	0.693	1.59 (0.18–13.89)	0.674
Maternal age at birth						0.92 (0.8–1.05)	0.22	0.93 (0.81–1.06)	0.275
Passive smoking						1.87 (0.08–44.98)	0.701	2.09 (0.09–51.35)	0.652
Maternal illness during pregnancy
Hypertension					1.69 (0.13–21.71)	0.687	1.74 (0.12–24.27)	0.682
Preeclmasia					4.49 (1.64–12.31)	0.004	4.47 (1.61–12.42)	0.004
Mental disorders
ADHD							6.2 (0.41–92.77)	0.186
Tic disorder							NA	NA
Conduct disorder							NA	NA

ADHD = attention deficit hyperactivity disorder, aOR = adjusted OR, CBCL = Child Behavior Checklist, CI = confidence interval, Ldn = day–night equivalent level, NS = noise sensitivity, OR = odds ratio.

Internalizing problems

Noise sensitivity was significantly associated with the clinical range of internalizing problems, and the adjusted OR (aOR) for a 1-point increase in noise sensitivity was 1.42 (aOR = 1.42; 95% CI: 1.15, 1.75). Internalizing problems were significantly associated with maternal hypertension during pregnancy (aOR = 5.75, CI: 1.22, 27.12) and preeclampsia (aOR = 3.83, CI: 1.64, 8.92) [Table 5].

Externalizing problems

Noise sensitivity was significantly associated with the clinical range of externalizing problems. The aOR for a 1-point increase in noise sensitivity was 1.24 (aOR = 1.24, CI: 1.02, 1.51). The association between age and externalizing problems was not significant after adjustment for sex, monthly income, maternal age at birth, passive smoking, and maternal illness. However, after adjustment for mental disorders, the association became significant (aOR = 1.35, CI: 1.02, 1.78) [Table 6].

Table 6

Associations between externalizing problems of the CBCL and noise-related variables

		Model 1		Model 2		Model 3		Model 4
		OR (95% CI)	P value	aOR (95% CI)	P value	aOR (95% CI)	P value	aOR (95% CI)	P value
Noise sensitivity		1.26 (1.04–1.53)	0.017	1.26 (1.04–1.53)	0.017	1.27 (1.05–1.55)	0.016	1.24 (1.02–1.51)	0.033
Noise level (Ldn)		1.03 (0.99–1.08)	0.154	1.03 (0.98–1.08)	0.21	1.03 (0.98–1.08)	0.271	1.03 (0.98–1.08)	0.247
Age				1.28 (0.98–1.66)	0.066	1.27 (0.98–1.66)	0.076	1.35 (1.02–1.78)	0.038
Sex				0.58 (0.26–1.33)	0.199	0.57 (0.25–1.31)	0.185	0.55 (0.24–1.29)	0.17
Monthly income				1.19 (0.44–3.24)	0.728	1.44 (0.48–4.32)	0.512	1.53 (0.5–4.72)	0.456
Premature birth						3.95 (1.01–15.52)	0.049	3.08 (0.71–13.41)	0.135
Maternal age at birth						0.98 (0.87–1.09)	0.686	0.98 (0.88–1.1)	0.756
Passive smoking						3.67 (0.25–54.13)	0.343	4.09 (0.26–63.59)	0.315
Martenal illness during pregnancy
Hypertension					0.82 (0.07–9.41)	0.874	0.93 (0.08–11.3)	0.954
Preeclmasia					2.32 (0.87–6.19)	0.094	2.31 (0.85–6.25)	0.099
Mental disorders
ADHD							10.07 (0.85–119.69)	0.068
Tic disorder							NA	NA
Conduct disorder							NA	NA

ADHD = attention deficit hyperactivity disorder, aOR = adjusted OR, CBCL = Child Behavior Checklist, CI = confidence interval, Ldn = day–night equivalent level, NS = noise sensitivity, OR = odds ratio.

Total behavior problems

Total behavioral problems were significantly associated with the noise level and noise sensitivity, with or without multistep adjustment. The aOR for a 1-point increase in noise sensitivity was 1.33 (aOR = 1.33, CI: 1.04, 1.70) and that of a 1-dB increase in the noise level was 1.08 (aOR = 1.08, CI: 1.01, 1.15). In addition to the noise-related variables, the only variable that was significantly associated with behavioral problems was preeclampsia during pregnancy (aOR = 4.47, CI: 1.61, 12.42) [Table 7].

Noise sensitivity and income

The tests of interaction between noise sensitivity and household income on behavioral problem in multivariate analyses were not statistically significant (P values ranging from 0.55 to 0.84 for internalizing problems, 0.19 to 0.36 for externalizing problems, and 0.48 to 0.78 for total behavior problems).

In Table 8, we present the results from the analyses of the association between behavioral problems and noise sensitivity, stratified by monthly income. The effect estimates were higher in the low-income group than the high-income group, in analyses on internalizing, externalizing, and total behavioral problems. In the high-income group, the association between externalizing problems and noise sensitivity disappeared. The high-income group lived in environments with higher noise levels, lived closer to roads, and had a higher maternal age at birth. Students in the low-income group were shorter in height, had a lower body weight, had parents with a lower education level, had higher exposure to passive smoke, had a higher percentage of fathers who were smokers, and spent longer time on smartphone/computer game use [Supplement Table 9].

Table 8

Associations between behavioral problems and noise sensitivity stratified by income

CBCL	Income	n	Crude OR (95% CI)	P value	Model 1, OR* (95% CI)	P value
Internalizing problem	Low	185	1.54 (1.08–2.18)	0.016	1.63 (1.11–2.39)	0.012
	High	570	1.33 (1.08–1.65)	0.008	1.37 (1.09–1.71)	0.006
Externalizing problem	Low	185	1.54 (1.03–2.32)	0.037	1.62 (1.04–2.52)	0.032
	High	570	1.18 (0.96–1.44)	0.111	1.19 (0.97–1.46)	0.091
Total problem	Low	185	1.48 (0.99–2.21)	0.057	1.55 (1–2.4)	0.05
	High	570	1.26 (0.99–1.6)	0.059	1.29 (1.01–1.66)	0.044

CBCL = Child Behavior Checklist, CI = confidence interval, OR = odds ratio.**Correspondence to Adjusted for age, sex, noise level (day–night equivalent level).

Table 9

Demographic, socioenvironmental, and developmental variables of all participants according to income

Variables	All		Low income			High income					P value
	n	Mean ± SD or %	n	Mean ± SD or %				n	Mean ± SD or %
Height	243	150.83 ± 11.39	52		147.98 ± 12.43		191			151.6 ± 11.00		0.042
Weight	243	44.29 ± 12.06	52		41.13 ± 12.22		191			45.15 ± 11.90		0.033
NS	900	3.32 ± 2.03	221		3.37 ± 1.98		684			3.30 ± 2.05		0.566
Noise level (Ldn)	762	56.82 ± 9.38	187		55.47 ± 10.98		575			57.26 ± 8.77		0.024
Educational level of father												<0.001
Less than high school	2	0.2%	2		0.9%		0			0.0%
High school diploma	177	20.0%	82		38.3%		95			14.1%
Associate’s degree	159	17.9%	48		22.4%		111			16.5%
Bachelor’s degrees	423	47.7%	72		33.6%		351			52.2%
More than Master’s degrees	126	14.2%	10		4.7%		116			17.2%
Educational level of mother												<0.001
Less than high school	2	0.2%	1		0.5%		1			0.1%
High school diploma	197	22.0%	85		39.7%		112			16.4%
Associate’s degree	202	22.5%	63		29.4%		139			20.4%
Bachelor’s degrees	427	47.6%	60		28.0%		367			53.7%
More than Master’s degrees	69	7.7%	5		2.3%		64			9.4%
Distance to nearest road												0.015
Neighboring road	195	21.6%	36		16.2%		159			23.3%
<50 m	215	23.8%	49		22.1%		166			24.3%
>50 m, ≤100 m	292	32.3%	79		35.6%		213			31.2%
>100 m, ≤500 m	187	20.7%	56		25.2%		131			19.2%
≥500 m	15	1.7%	2		0.9%		13			1.9%
Father’s age at birth (years)	885	32.75 ± 3.84	210		32.41 ± 4.05		675			32.86 ± 3.77		0.139
Mother’s age at birth (years)	896	30.26 ± 3.74	214		29.70 ± 4.05		682			30.44 ± 3.62		0.012
Breast feeding												0.488
No	330	36.5%	76		34.5%		254			37.1%
Yes	574	63.5%	144		65.5%		430			62.9%
Smoking status of father												0.009
Never smoker	337	37.3%	65		29.8%		272			39.7%
Former smoker	244	27.0%	63		28.9%		181			26.4%
Current smoker	322	35.7%	90		41.3%		232			33.9%
Smoking status of mother												0.06
Never smoker	892	98.8%	215		97.7%		677			99.1%
Former smoker	6	0.7%	2		0.9%		4			0.6%
Current smoker	5	0.6%	3		1.4%		2			0.3%
Passive smoking
No	779	86.3%	177		80.8%		602			88.0%		0.007
Yes	124	13.7%	42		19.2%		82			12.0%
Smartphone usage												0.002
Never	115	12.7%	27		12.2%		88			12.9%
<1 h	374	41.4%	66		29.9%		308			45.1%
1–3 h	306	33.8%	96		43.4%		210			30.7%
3–5 h	90	10.0%	23		10.4%		67			9.8%
≥5 h	19	2.1%	9		4.1%		10			1.5%
Computer/video game usage												<0.001
Never	100	11.1%	17		7.7%		83			12.2%
<1 h	462	51.3%	87		39.4%		375			55.1%
1–3 h	289	32.1%	96		43.4%		193			28.4%
3–5 h	42	4.7%	17		7.7%		25			3.7%
≥5 h	8	0.9%	4		1.8%		4			0.6%

CI = confidence interval, Ldn = day–night equivalent level, NS = noise sensitivity, OR = odds ratio.

Sensitivity analysis

In the sensitivity analysis, the significance of the associations among noise sensitivity, noise levels, and behavioral problems was maintained after excluding those students who had lived in their respective residential area for less than 1 year [n = 20, Supplement Table 10].

Table 10

Association among behavioral problems, noise, and noise-sensitivity, excluding participants living in their residential area for less than 1 year

	Crude		Adjusted I		Adjusted II		Adjusted III
	OR (95% CI)	P value	aOR (95% CI)	P value	aOR (95% CI)	P value	aOR (95% CI)	P value
Internalizing problem
Noise sensitivity	1.41 (1.17–1.71)	<0.001	1.42 (1.17–1.73)	<0.001	1.45 (1.18–1.79)	0.001	1.42 (1.15–1.75)	0.001
Noise level (Ldn)	1.01 (0.97–1.05)	0.701	1.01 (0.97–1.06)	0.564	1.02 (0.97–1.06)	0.432	1.02 (0.97–1.06)	0.457
Externalizing problem
Noise sensitivity	1.26 (1.04–1.53)	0.018	1.26 (1.04–1.53)	0.017	1.27 (1.05–1.55)	0.016	1.24 (1.02–1.51)	0.032
Noise level (Ldn)	1.03 (0.99–1.08)	0.156	1.03 (0.98–1.08)	0.217	1.03 (0.98–1.08)	0.284	1.03 (0.98–1.08)	0.250
Total problem
Noise sensitivity	1.36 (1.08–1.71)	0.008	1.36 (1.08–1.7)	0.008	1.35 (1.06–1.72)	0.014	1.32 (1.04–1.68)	0.023
Noise level (Ldn)	1.07 (1.01–1.14)	0.024	1.08 (1.01–1.15)	0.026	1.07 (1–1.14)	0.037	1.07 (1–10.15)	0.037

aOR = adjusted OR, CI = confidence interval, Ldn = day–night equivalent level, NS = noise sensitivity, OR = odds ratio. Sensitivity analysis: participants who had been living at their current house less than 1 year were excluded. Adjusted I: age, sex, income. Adjusted II: premature birth, maternal age at birth, maternal disease during pregnancy (hypertension, preeclampsia), passive smoking. Adjusted III: mental disorders (ADHD, tic disorder, conduct disorder).

Discussion

This population-based study investigated associations between noise-related variables and the mental health of children and adolescents, measured using the CBCL. Noise was significantly associated with total behavioral problems, and noise sensitivity was significantly associated with internalizing, externalizing, and total behavioral problems after adjustment for various variables that could affect the children and adolescents’ behavioral problems. This is the first report on the impact of noise sensitivity in children and adolescents, and our findings suggest that the negative impact of noise and noise sensitivity on mental health occurs from childhood.

Road-traffic noise in residential areas was significantly associated with total behavioral problems but not with internalizing and externalizing problems. In a longitudinal study on German children, road traffic noise and noise caused by neighbors were found to be risk factors for high total difficulty scores as well as high scores on the emotional symptoms, conduct problems, and hyperactivity subscales of the Strengths and Difficulties Questionnaire.[43] However, this study was limited by the fact that it was based on subjectively assessed noise, not actual noise levels. Previous studies on the effect of road traffic noise exposure at school have reported no link between road traffic noise at school and children and adolescents’ mental health.[374445] Meanwhile, aircraft noise exposure at school was associated with more hyperactivity symptoms.[364445] Therefore, the effects of noise on children could vary depending on the type of noise (e.g., road traffic, air traffic, neighborhood noise) and the exposure location (e.g., school, home). In an additional analysis of our study, the proportion of students with behavioral problems did not differ according to the noise level at their school. Altogether, noise exposure in the evening and at night in residential areas may be more important for mental health than that during school activities.

In the present study, a 1-point increase in the noise sensitivity of children and adolescents increased their internalizing problem score by 1.42, their externalizing problem score by 1.24, and their total behavioral problem score by 1.33. This result is in concordance with previous studies on adults that have reported a relationship among noise sensitivity and depression, anxiety, and nonspecific somatic complaints.[294647] It has been reported that children with internalizing problems are more vulnerable to depression later on[484950] and that those with externalizing problems are more likely to be diagnosed with anxiety disorder in the future.[5051] Behavioral problems in children and adolescents could lead to impairments in social skills, self-confidence, and peer relationships.[52] As a result, these behavioral problems might have long-term implications for educational attainment and occupational opportunities.[3940] Therefore, it is important to screen such problems early and intervene to avoid negative outcomes; noise sensitivity may be an important indicator in this process.

In the present study, the association between children and adolescents’ behavioral problems and noise sensitivity was more consistent than the association with noise levels. Unlike noise, which was associated only with total behavioral problems, noise sensitivity showed significant associations with more specific items such as internalizing and externalizing problems, and the significance level of these associations was lower for noise sensitivity than for noise. For total behavioral problems, the effect of a 1-point increase in noise sensitivity was comparable to a 5-dB increase in noise level (OR = 1.40). Meanwhile, noise levels were not significantly different between the groups with high and low noise sensitivity, suggesting that noise sensitivity affects behavioral problems by a mechanism that is different from noise. This result corroborates a previous study conducted in the same region by our research group that showed that noise sensitivity rather than noise was predictive of adults’ mental health, as manifested in depression, anxiety, insomnia, and stress.[53] It is also consistent with the findings of previous studies showing that noise sensitivity, independent of the noise level, is a factor that negatively influences physical and mental health.[26295455] As higher cortisol levels[56] and hyperactivation of the sympathetic–adrenal–medullary system and hypothalamic–pituitary–adrenal axis[575859] are observed in individuals with high noise sensitivity, it is assumed that stress reactivity is increased in those individuals.

Because some studies have suggested that prolonged noise exposure increases noise sensitivity,[5560] when interpreting the results of the present study, it should be considered that the exposure duration was not included in the analysis. However, the significance of the association between noise sensitivity and behavioral problems in children and adolescents did not change after excluding people who had lived for less than 1 year in their respective residential area.

In this study, gestational hypertension was significantly associated with internalizing problems, and preeclampsia was significantly associated with internalizing and total behavioral problems. Gestational hypertension and preeclampsia can lead to abnormalities in blood perfusion and fetal nutrition. It can thus be assumed that both would have a negative effect on brain development, which would increase behavioral problems as well.[6162] In a previous study that investigated the relationships between gestational hypertension or preeclampsia and behavioral problems, gestational hypertension had a positive association with behavioral problems, whereas preeclampsia had a negative association with behavioral problems in some age groups,[63] suggesting that the use of maternal antihypertensive agents during pregnancy may have affected the neurocognitive functions of the children.[6465] However, information on the use of antihypertensive agents was not obtained in this study.

In genetic studies in monozygotic and dizygotic twins, noise sensitivity showed a heritability of 40%.[66] Because noise sensitivity played a role as an indicator of mental health from childhood in the present study, it could be a trait that is inherited and expressed from childhood. Some studies have insisted that noise sensitivity is a kind of personality trait, and studies investigating relationships between personality traits and noise sensitivity show that neuroticism has a relatively consistent association with noise sensitivity.[2167686970] However, there have been controversies regarding the association between noise sensitivity and personality, and other factors besides personality seem to be involved in the variability of noise sensitivity.[71] In addition, considering that personality is being formed during childhood and the adolescent period, longitudinal studies are needed to investigate the time course of noise sensitivity and whether it is maintained during development.

We found that the magnitude of the association between noise sensitivity and behavioral problems decreased in the high-income group. A high income level seems to be a protective factor against the negative impact of noise and noise sensitivity, in that the influence of noise sensitivity on behavioral problems was found to be reduced even when the average noise level was higher in the high-income group. In particular, the significance of the association between noise sensitivity and externalizing problems disappeared in the high-income group. This result is consistent with previous studies reporting that household income is more closely related to externalizing problems than to internalizing problems.[4727374] Furthermore, the present findings that the association between noise sensitivity and internalizing problems are maintained in the high-income group further support the claim mentioned earlier that noise sensitivity is a kind of personality trait. However, the noise used in this study was measured on the exterior walls of the building. Therefore, the negative impact of noise may have been reduced by living in homes with better sound insulation in high-income groups.

Children and adolescents from low-income families showed a higher magnitude of association between noise sensitivity and behavioral problems than those from high-income families. Children and adolescents from low-income households have high scores on both self-reported[75] and biologically assessed stress.[76] It can thus be assumed that children and adolescents with high noise sensitivity (individuals who are vulnerable to environmental stress factors, such as noise) become more vulnerable to stress because of their external environment (at a low socioeconomic level). In contrast, children in low-income families are more likely to experience multiple physical and psychosocial stressors which have an impact on mental health.[77] These stressors could affect the association between noise-related variables and behavioral problem.Effortful control is a regulative temperament factor and defined as “the ability to inhibit a dominant response to perform a subdominant response.”[78] Children with high level of effortful control could modulate their negative emotions well and have less psychopathology.[79] Effortful control consists of attentional control, which manifests as an internalizing problem, and inhibitory control, which manifest as an externalizing problem.[80] Noise sensitivity was more consistently related to the internalizing problem in this study. Therefore, in the temperament aspect, noise sensitivity could be more linked to a lack of attention control rather than inhibitory control. Furthermore, it is assumed that children with high noise sensitivity may have been compromised in their ability to focus and shift attention from noise as needed.

The higher rate of passive smoking and higher smartphone and game use in the low-income group [Supplementary Table 9] could be due to an environmental effect of low socioeconomic status, such as low accessibility to health knowledge because of poor parental education; however, passive smoking or addictive behaviors could also be a biological effect on brain development. It has been assumed that the auditory cortex, brainstem, reticular formation, amygdala, and hippocampus are involved in noise-induced responses.[81] In a study using electroencephalography (EEG) and functional magnetic resonance imaging (fMRI),[82] an attempt was made to find differences in physical sensitivity according to noise sensitivity, but no definite conclusion could be reached. Further studies are therefore needed to understand the biological mechanism of noise sensitivity.

Strengths and Limitations

As a large population-based study conducted in two large cities in Korea, the present study confirms the association between noise and noise sensitivity and behavioral problems in children and adolescents. The main strength of our study is that the noise level was measured accurately, because we created a noise map using direct measurements, unlike previous studies that used subjective noise estimates or outdated existing noise maps. In addition, this is the first study that examines the effects of noise sensitivity on the mental health of children and adolescents.

However, the current study has several limitations. First, the measurements of noise sensitivity reflect not only the evaluation of the participant but also the opinions of their parents. Thus, there may be a controversy as to whether findings related to noise sensitivity from previous studies in adults can be applied to the interpretation of our results in children and adolescents. However, parental reporting is an important source of information when studying children and adolescents, because these participants have limited self-reflection. In addition, as the scale for assessing behavioral problems used in this study is also based on parental reporting, the fact that noise sensitivity also reflects parental evaluations is consistent and seems appropriate.

Second, in this study, only the residents of two large cities, Seoul and Ulsan, were included. Generally, large cities have higher average noise levels than do rural areas, and the proportion of aircraft noise and road traffic noise is higher, resulting in different types of noise distribution compared with rural areas. Urban areas also reflect different socioeconomic factors. Therefore, the effects of noise and noise sensitivity found in this study might not apply to the general population but only to specific groups.

Third, the noise level was calculated based on the exterior wall of the building. Therefore, the association between noise and mental health in this study may not reflect the effect of indoor noise actually exposed by the participants in the building.

Fourth, for there was no significant interaction between noise sensitivity and income in analyses with behavioral problems, care should be taken not to overestimate the differences between high- and low-income groups in stratified analyses.

Fifth, because this is a cross-sectional study, we could not determine the causality of the association. To investigate temporal changes in noise sensitivity in children and adolescents as well as to infer causality, further studies are needed.

Conclusion

We observed the negative mental health effects of noise in childhood. In addition, noise sensitivity and noise itself play an important role in the mental health of children and adolescents. The associations between noise sensitivity and emotions or behaviors were stronger in the low-income group. Therefore, when developing strategies to cope with mental health problems caused by noise in children and adolescents, subgroups classified by noise sensitivity or the socioeconomic status may require different approaches. Noise is an inevitable problem in modern societies, and considering long-term implications on behavioral problems in children and adolescents in terms of their educational and professional prospects, additional studies, including longitudinal studies, are needed.

Financial support and sponsorship

This study was supported by the Korea Ministry of Environment (MOE) as The Environmental Health Action Program (grant number: 2014001350001). The sponsor had no further role in designing and carrying out the study, as well as in interpretation of the data, drafting of the manuscript, and decision to submit for a publication.

Conflicts of interest

There are no conflicts of interest.