Mental health effects of exposure to environmental noise at home: A systematic review of potential mediating pathways

Abstract

Objective:

Research has established a mental health impact of exposure to environmental noise, but specific mechanisms driving this association are poorly understood. Several plausible mediating factors have been proposed, including noise annoyance and sleep disturbance. We undertook a systematic review of the literature to synthesize the existing evidence on possible mediating pathways between exposure to environmental noise and poor mental health in adults.

Methods:

We systematically searched Medline, PsycINFO, Embase, and Web of Science databases to identify relevant studies published up to May 2023. Studies of the mental health impact in adults of environmental noise at home that examined at least one mediator were included. Included studies were assessed for risk of bias. Findings from included studies were synthesized narratively. Study heterogeneity and high risk of bias precluded meta-analysis.

Results:

Of the 892 studies retrieved, 14 met the inclusion criteria. These spanned 13 unique mediators. Some studies examined multiple mediators. The most commonly examined mediators were noise annoyance (nine studies), sleep disturbance (four studies), and physical activity (two studies). Taken together, these studies provide support for noise annoyance and sleep playing a role in the mediation of the noise-mental health association, while evidence was limited or absent for other mediators. The small number of studies and high risk of bias leads to low certainty of evidence.

Conclusions:

Noise annoyance and sleep disturbance are likely mediators of the relationship between exposure to environmental noise at home and poor mental health. However, higher quality and longitudinal researches are needed to clarify these and other potential mediating pathways.

KEY MESSAGES

• (1)Noise annoyance and sleep disturbance are the most commonly studied potential mediators of the relationship between exposure to environmental noise and poor mental health.
• (2)In this systematic review, we found that the existing evidence lends some support to noise annoyance and sleep as likely important mediators, but the number of published studies is small, and study quality is generally low.
• (3)Higher quality and longitudinal research is needed to clarify these and other potential mediating pathways.

INTRODUCTION

Environmental noise is an increasingly pervasive global public health issue. As a consequence of rapid urbanization, which will see the proportion of the global population living in urban areas increase from 55% in 2018 to 68% in 2050,[1] a growing number of people are being exposed to potentially harmful levels of noise. In Europe alone, an estimated 1 million healthy years of life are lost annually because of the negative health effects associated with exposure to environmental noise.[2]

Noise is the presence of unwanted sound.[3] Important sources of environmental noise include transportation, such as road, rail, air traffic, construction, and industry, as well as amplified music and wind turbines.[4] Research has established a relationship between exposure to environmental noise and several physical and mental health outcomes. Exposure to noise levels above 40–54 decibels (dB) over the entire day (L_den) is associated with cardiovascular disease, cognitive impairment, tinnitus, hearing loss, and poor mental health, among other outcomes.[5,6,7] Regarding mental health specifically, a positive association between exposure to environmental noise and both depression and anxiety has been established. A meta-analysis by Dzhambov and Lercher,[8] for instance, found that exposure to road traffic noise was associated with 4% and 12% higher odds of depression and anxiety, respectively, per 10 dB increase in L_den. Other research has been in broad agreement, albeit with differing estimated magnitudes of effect.[9] However, despite there being evidence of an association between environmental noise and mental health, the underlying drivers of the association are not well understood.[10]

Several mediating factors have been proposed to explain the association between environmental noise and mental health [Figure 1]. These include emotional responses (e.g., annoyance) and behavioral changes (e.g., reduced physical activity, sleep disturbance, increased alcohol consumption),[11,12] for which there is evidence of an association with environmental noise exposure.[5,13,14,15] Multiple mediators may be involved, either in parallel or sequentially. For example, noise may elicit feelings of annoyance, characterized by irritation and discomfort, leading to a subsequent decline in mental health, while noise may also independently affect mental health via sleep disturbance (parallel mediation pathways). Conversely, feelings of noise annoyance may disturb sleep and, through sleep, affect mental health (sequential mediation). In the case of both annoyance and sleep, the pathophysiological mechanisms underlying noise-induced mental health consequences are believed to occur via stress response mechanisms involving dysregulation of neurological and endocrine pathways and systemic inflammation–known biological risk factors for depression and anxiety when exposure is prolonged.[10] Other potential behavioral mediators may arise as maladaptive coping responses to chronic stress or via different biological mechanisms, such as impaired cognitive processes.[10] Better understanding the direct and indirect pathways from environmental noise to mental health will assist in the development of interventions to protect populations from harmful exposure (e.g., source control interventions such as restricting air traffic above residential areas at nighttime; residential design and construction interventions such as providing subsidies for double glazing of windows in homes affected by noise from traffic or industry; and behavior change interventions to avoid or reduce exposure).[4]

Figure 1.

Overview of proposed direct and indirect (mediated) pathways from environmental noise exposure to mental health outcomes.

Accordingly, we undertook a systematic review of the literature to answer the question, “What factors mediate the association between environmental noise at home and mental health outcomes in adults?” To our knowledge, this review is the first to systematically synthesize the existing epidemiological evidence for putative mediating pathways that may explain the association between environmental noise exposure at home and mental health. Our review focuses on epidemiological studies of mediating pathways rather than clinical studies of biological mechanisms. The focus herein on the home environment reflects an acknowledgment that in many parts of the world, people spend more than 80% of their time inside their homes.[16] Thus, the home and its surrounding environment are important sites for public health intervention. By addressing this gap in the literature, we thereby aim to both identify where evidence is lacking and to provide insights for future research, policy and practice in this realm.

MATERIALS AND METHODS

Registration and Protocol

As per Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines,[17] our review protocol was registered in the International Prospective Register of Systematic Reviews (PROSPERO) with a register number of CRD42023416773.

Eligibility Criteria

The eligibility criteria are summarised below, and full details are provided in Supplementary Table S1.

Supplementary Table S1.

Eligibility criteria

	Include	Exclude
Population	Community dwelling populations comprising at least 90% adults aged 18 and above	Individuals living in care facilities such as aged care facilities or hospitals Study populations selected due to presence of outcome Non-human populations Studies where less than 90% of population is aged 18 and above
Exposure	Environmental noise measured in the home	Noise exposure measured out of the home Exposure to noise being actively created by participant e.g. listening to music
Outcome	Individual mental health reported using objective measurement tools or self-reported
Study Designs	Studies investigating mediating factors of the association between environmental noise and mental health	Reviews

Study Characteristics

Studies were included if they investigated any mediating factors of the association between environmental noise and mental health or well-being. All study designs except literature reviews were eligible for inclusion in this review.

Population

The review focussed on adults living in the community. Studies were included if more than 90% of participants were 18 years or older or if results were available separately for adults and children. There was no upper limit for the age of participants. Studies looking at clinical, pediatric, and nonhuman populations were excluded. Studies that looked at individuals who were living in residential care, such as aged care homes or inpatients in hospitals, were excluded due to differences in environmental exposures and baseline health in these settings compared to community environments. Children were excluded due to differences in how adults and children experience and respond to environmental exposures, both physiologically and psychologically.[18,19,20]

Exposure

The primary focus of this review was on exposure to environmental noise at home (whether from an internal or external source, but excluding noise created by the study participant, such as listening to music). We define home as the place where a person lives regularly. In most cases, this was houses, units, or apartments, though some studies included dorm rooms in residential colleges or similar. Exposure to noise at home can be measured in a variety of ways, directly or indirectly and objectively or subjectively, and we did not exclude studies on this basis. Rather, the key criterion for inclusion was that the exposure measure was intended to capture the experience of noise by study participants whilst indoors at their residences.

Outcomes

The outcome of interest was mental health or well-being, including both general mental health or symptoms or diagnosis of a mental health condition. Eligible studies used either self-reported or clinically measured mental health outcomes or both. Studies that reported health-related quality of life (HRQoL) were also included, where the component measure for mental health was reported separately from physical health components.

Report Characteristics

Only peer-reviewed articles available in English were eligible for inclusion. No restrictions were placed on the date of publication.

Search Strategy

We searched four online scientific databases: Ovid MEDLINE (1946 to present), PsycINFO (1806 to present), Embase (1947 to present), and Web of Science (1900 to present). The search was conducted on 31st March 2023. The search string combined terms for environmental noise exposure, mental health, and mediation. We also undertook backward citation searching of relevant existing reviews to identify any further studies that may be eligible for inclusion, which were then located and introduced into the full-text screening. Exclusions based on population age were made during the screening stages. The full search strategy for each database is outlined in Supplementary Table S2.

Supplementary Tables S2a & S2b.

Database search strategies, We searched Web of Science, Embase, Medline and PsycInfo databases for records containing combinations of the following terms for environmental noise, mental health and mediation, in any field:

Topic	Search terms
Environmental noise	"environmental noise*" OR "background noise*" or "ambient noise*" OR "noise pollution" OR "incidental noise*" OR "surrounding noise*" OR "environmental sound*" OR "background sound*" OR "ambient sound*" OR "soundscape" OR "sound pollution" OR "incidental sound*" OR "surrounding sound*” OR “unwanted noise”
	AND
Mental health and wellbeing	"mental health" OR "psychological health" OR "emotional health" OR "mental well*" OR "psychological well*" OR "emotional well*" OR depress* OR anxi* OR stress
	AND
Mediating pathways	mediat* OR “indirect effect*” OR pathway* OR attenua* OR mechanism OR intermediate
Search results for each database are show below:
Database	Results
Web of Science	473
Embase	235
Medline	135
PsycInfo	49

Screening Process

Search results were imported into Covidence systematic review software (Veritas Health Innovation, Melbourne, Australia, www.covidence.org), and duplicates were removed automatically. All remaining abstracts were screened independently by four reviewers (KG, ES, KM, and AM) within Covidence, and conflicts were resolved through discussion between the reviewers who had voted. Full texts were also screened independently by four reviewers (KG, ES, KM, and AM) within Covidence. For consistency, one reviewer (KG) screened all studies in both the abstract and full-text screening phases. Conflicts regarding whether to include or exclude the paper were resolved through a combination of a discussion between the reviewers who had voted or the input of three reviewers (KM, AM, ES). Conflicts of exclusion reasons were resolved by a single reviewer (KG) using an ordered hierarchy of exclusion reasons after ensuring both conflicting responses were relevant. The ordered hierarchy was as follows: full text not available, full text not available in English, review, wrong exposure/outcome, wrong setting, pediatric population, no mediation analysis.

Data Extraction

Information collected included summary statistics related to the study population, the date and location of the study, the study design, the exposure, and the outcome, including measures used, the mediators studied, and the statistical techniques used for analyzing mediation. The form was created within Covidence by a single reviewer (KG) and checked for usability and completeness by a second reviewer (KM). The form was created iteratively whilst being tested against the studies from which data extraction was required. All studies were extracted by a single reviewer (KG), with two reviewers (ES, KM) cross-checking all extractions against the original sources. Where the second reviewer was unable to determine if a discrepancy was present, then a third reviewer (KM) was consulted.

Risk of Bias Assessment

Existing validated risk-of-bias tools do not assess the risk of bias specifically relating to mediation analysis. Potential sources of bias in observational mediation analyses—additional to those relevant for all observational studies—include mismeasurement of mediator, unmeasured confounding of the exposure-mediator or mediator-outcome pathways, and failure to check or use methods to account for exposure-mediator interaction. We, therefore, assessed the risk of bias in each study using a tailored assessment tool developed and used elsewhere for observational studies of mediation.[21] Two reviewers (KM and AM) independently assessed each study and jointly resolved discrepancies.

Synthesis Methods

Studies were synthesized according to the mediators analyzed, with studies that investigated multiple mediators being included in all relevant mediator subgroups. While a meta-analysis was considered when writing the protocol, it was later determined that meta-analysis was not feasible due to heterogeneity in exposure and outcome measurement methods, statistical approaches, and the small number of studies available for each mediator.

For each mediator examined in more than one study, we first reviewed studies that included a single mediation model for only the mediator of interest. Where relevant, we then additionally synthesized results from models, including the mediator of interest in parallel or sequence with other proposed mediators.

Synthesis was not possible for mediators that were only included in a single study; therefore, we simply report the evidence and authors’ conclusions about each such mediator. We also considered the authors’ rationale for investigating the proposed mediator and discuss this in light of other research to conclude whether existing evidence supports further studies examining the same proposed mediator.

Certainty Assessment

The Grading of Recommendations Assessment, Development and Evaluation (GRADE) tool was used to determine the level of certainty of the evidence in this review.[22] The GRADE tool assesses study design, risk of bias, inconsistency, indirectness, imprecision, and publication bias to establish the overall quality of the evidence. The assessment was completed independently by two reviewers (KM and AM). The results of the GRADE assessment are reported in the analysis to provide readers with a clear understanding of the strength of the evidence underpinning the conclusions of the review.

RESULTS

Study Selection

The PRISMA flow diagram in Figure 2 outlines the screening process. Our search returned a total of 892 records. After removing duplicates, 615 records were included in the title and abstract screening, at which stage 550 records were excluded. In addition to the 65 records remaining after title and abstract screening, an additional 24 records were identified for full-text screening from backward citation searching of relevant reviews. Of the 89 full texts screened, 75 were excluded, leaving a total of 14 for inclusion in this review.

Figure 2.

PRISMA flow diagram of literature search and screening process. Note: PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses; Diagram generated using Covidence software.

Study Characteristics

Most of the included studies were conducted in European countries (10 studies; 3 Switzerland, 2 Bulgaria, 2 The Netherlands, 1 Denmark, 1 England, 1 Wales) while the remainder (4 studies) were conducted in Nepal (1 study), Canada (1 study) and the USA (2 studies). Ten of the studies were published in the last five years (2018 or later). Ten of the studies used a cross-sectional study design, while the remaining four were cohort studies. Table 1 displays an overview of the included studies; more detailed information is in Supplementary Tables S3–S5.

Table 1.

Summary Characteristics of Studies Included in the Systematic Review.

Study	Population	Sample Size	Setting	Mediators Considered	Study Design
Bakker et al. [23]	Adults living within 2.5 km of a wind turbine	725	The Netherlands	Noise annoyance, sleep disturbance	Cross-sectional
Oiamo et al. [24]	Adults	603	Ontario, Canada	Noise annoyance	Cross-sectional
Roswall et al.[25]	Adults aged 50–64	45,271	Copenhagen, Denmark	Waist circumference, smoking status, alcohol intake	Prospective cohort
Peltz and Rogge[26]	University students living in a residential college	335	New York, USA	Sleep disturbance	Cross-sectional
Dzhambov et al.[12]	Medical students	720	Plovdiv, Bulgaria	Noise annoyance, sleep disturbance, neighborhood restorative quality, neighborhood social cohesion, physical activity	Cross-sectional
Dzhambov et al.[27]	Medical students	437	Plovdiv, Bulgaria	Noise annoyance	Cross-sectional
Klompmaker et al.[28]	Adults	354,827	The Netherlands	Chronic physical disease	Cross-sectional
Cerletti et al.[29]	Adults	2035	Switzerland	Noise annoyance	Prospective cohort
Eze et al.[11]	Adults	4581	Switzerland	Noise annoyance; BMI; physical activity	Prospective cohort
Nazneen et al.[30]	Local residents, students (aged 15+), police officers, shopkeepers	2500	Peshawar, Pakistan	Headaches	Cross-sectional
Peltz[31]	Undergraduate students	257	USA	Restedness	Cross-sectional
Stansfeld et al.[32]	45–59-year-old males	1320	Caerphilly, Wales	Noise annoyance	Prospective cohort
Fisher et al.[33]	Adults	282	Leeds, UK	Biodiversity: actual, perceived	Cross-sectional
Gomm and Bernauer[34]	Adults	5729	Switzerland	Perceived noise exposure	Cross-sectional

Note: BMI, body mass index.

Supplementary Table S3.

Key data extracted from included studies

Study	Study design and setting	Environmental Noise Exposure	Mental Health Outcome(s)	Mediator 1	Mediator 2	Mediator 3	Mediator 4	Mediator 5	Statistical method used to assess mediation
Bakker et al (2012)1	Cross sectional study (N=725) Netherlands	Wind turbine noise (objective)	Psychological distress: General health Questionnaire (GHQ)-12	Noise annoyance: selfreported 11-pt scale for each of indoors and outdoors	Sleep disturbance: selfreported 5-pt scale				SEM (LISREL)
Cerletti et al (2020)2	Cohort study (N=2035) Switzerland	Source-specific transportation noise (objective)	Health Related Quality of Life (HRQoL) (mental health): Short Form (SF)-36 questionnaire (mental health domain)	Noise annoyance: selfreported 11-pt scale					Predictive multiple quantile regression (in STATA)
Dzhambov et al (2018)3	Cross sectional study (N=720) Bulgaria	Residential noise (objective)	Mental ill-health: GHQ-12 [excluding item on loss of sleep due to worry]	Noise annoyance: 4 survey items (weighted mean)	Sleep disturbance:   2 survey items (mean)	Neighbourhood restorative quality: mean of 2 survey items from Perceived Restorativeness Scale	Neighbourhood social cohesion: mean of 3 items from the Perceived Neighbourhood Social Cohesion questionnaire	Physical activity: daily energy expenditure in METs, derived from self-reported SQUASH tool	SEM
Dzhambov et al (2019)4	Cross sectional study (N=437) Bulgaria	Road traffic Noise (objective)	Depression: Patient Health Questionnaire (PHQ)-9 Anxiety: Generalised Anxiety Disorder (GAD)-7 questionnaire	Noise annoyance: selfreported 5-pt scale					Product-of-coefficients approach (implemented with PROCESS macro in SPSS)
Eze et al (2020)5	Cohort study (N=4581) Switzerland	Railway, road and aircraft noise (objective)	Depression: SF-36 Mental Component Summary score	Noise annoyance: selfreported 11-pt scale	Body Mass Index: objectively measured	Physical activity: selfreported mins per week of MVPA (dichotomised)			Generalised SEM (in STATA)
Fisher et al (2022)6	Cross sectional study (N=282) England	Noise levels (objective)	Mental wellbeing: Short Warwick-Edinburgh Mental Wellbeing Scale	Actual biodiversity: transect sampling (tree, pollinator and flowering plant richness)	Perceived biodiversity: estimated number of species (trees, pollinators and flowering plants)				SEM (in R)
Gomm & Bernauer (2023)7	Cross sectional study (N=5729) Switzerland	Traffic noise (objective)	General mental health: GHQ-12	Perceived noise exposure: self-reported 5-pt scales (mean: traffic; industry)					Product-of-coefficients approach (implemented with PROCESS macro in R)
Klompmaker et al (2019)8	Cross sectional study (N=354,827) Netherlands	Road traffic noise (objective)	Mental Health: Kessler-10 scale Psychiatric medication use: prescription data	Chronic physical disease: self-reported physiciandiagnosis					Causal mediation analysis (implemented using 'mediation' package in R)
Nazneen et al (2020)9	Cross sectional study (N=2500) Pakistan	Noise levels (objective)	Subjective health complaints and psychological symptoms: survey	Headaches:   self-reported 6-pt scale of frequency					SEM (in SPSS)
Oiamo et al (2015)10	Cross sectional study (N=603) Canada	Road traffic noise (objective)	HRQoL (mental health): SF-12 mental health domain and Mental Component Summary score	Noise annoyance: selfreported 11-pt scale					SEM (in Mplus)
Peltz & Rogge (2016)11	Cross sectional study (N=335) USA	Environmental noise disturbances (subjective)	Depressive symptoms: PHQ-9 (excluding 2 sleep items)	Sleep disturbance: Pittsburgh Sleep Quality Index					SEM (in Mplus) & MacKinnon’s Asymmetric Confidence Interval approach (using 'mediation' package in R)
Peltz (2020)12	Cross sectional study (N=257) USA	Nighttime noise disturbance (subjective)	Depressive and anxiety symptoms: adapted PHQ-4	Restedness:   self-reported 11-pt scale of feeling upon awakening					Multilevel SEM models (Mplus)
Roswall et al (2015)13	Cohort (N=45,271) Denmark	Road traffic noise (objective)	HRQoL (mental health): SF-36 Mental Component Summary score	Waist circumference: self-measured	Smoking status: selfreported	Alcohol intake:   self-reported			Linear regression +/- adjustment for lifestyle factors
Stansfeld et al (2021)14	Cohort study (N=1320) Wales	Road traffic noise (objective)	Psychological ill-health: GHQ-30	Noise annoyance: selfreported 5-t scale					Cox Proportional Hazards regression +/- adjustment for noise annoyance

Supplementary Table S5.

Mediation pathways examined and key results

Study	Mediation pathways examined	Key results
Bakker et al (2012)	Wind Turbine Noise – Noise Annoyance – Psychological Distress Wind Turbine Noise – Sleep Disturbance – Psychological Distress Wind Turbine Noise – Noise Annoyance – Sleep Disturbance – Psychological Distress	Among respondents who reported that they noticed wind turbine noise, exposure was correlated with annoyance (r = 0.27, p<0.05) which in turn was correlated with psychological distress (r = 0.17, p<0.05). No evidence of mediation by noise annoyance for other subgroups. Sleep was not a mediator other than through noise annoyance.
Cerletti et al (2020)	Transportation Noise – Noise Annoyance – HRQoL/mental health	No evidence of mediation observed. Associations between source-specific transportation noise exposure and HRQoL remained stable after adjusting for noise annoyance (and for the mental health domain of HRQoL specifically, associations were non-significant)
Dzhambov et al (2018)	Residential Noise – Noise Annoyance – Mental Health Residential Noise – Noise Annoyance – Restorative Quality – Social Cohesion – Mental Health Residential Noise – Noise Annoyance – Restorative Quality – Physical Activity – Mental Health Residential Noise – Noise Annoyance – Restorative Quality –Mental Health Residential Noise – Noise Annoyance – Social Cohesion – Mental Health Residential Noise – Noise Annoyance – Sleep – Mental Health Residential Noise – Noise Annoyance – Physical Activity – Mental Health	Standardised total indirect effect of noise on mental health = 0.035 (95% CI: 0.010, 0.065), p=0.004). Explaining this, sequential mediation was observed via two pathways: (1) noise exposure - noise annoyance - restorative quality - physical activity - mental health (p=0.010); and (2) noise exposure - noise annoyance - sleep - mental health (p=0.003). No significant evidence of mediation was found for other paths.
Dzhambov et al (2019)	Traffic Noise – Noise Annoyance – Depression Traffic Noise – Noise Annoyance – Anxiety	Evidence of mediation by noise annoyance. Indirect effect on depression: unstandardised linear regression coefficient = 0.30 (95% CI: 0.01, 0.71) p<0.05; OR=1.20 (1.03, 1.50). Indirect effect on anxiety: unstandardised linear regression coefficient = 0.35 (95% CI: 0.06, 0.76) p<0.05; OR=1.20 (1.01, 1.52). Noise sensitivity moderated the relationship such that indirect effects through noise annoyance were limited to individuals with low noise sensitivity.
Eze et al (2020)	Road Traffic Noise – Noise Annoyance – Depression Road Traffic Noise – Physical Activity – Depression Road Traffic Noise – BMI – Depression Railway Noise – Noise Annoyance – Depression Railway Noise – Physical Activity – Depression Railway Noise – BMI – Depression Aircraft Noise – Noise Annoyance – Depression Aircraft Noise – Physical Activity – Depression Aircraft Noise – BMI – Depression	Noise annoyance mediated the association between both road traffic noise and aircraft noise (but not railway noise) and depression. No evidence of mediation by BMI or physical activity. Standardised indirect effects on depression via noise annoyance: road traffic noise β=0.017 (95%CI 0.005, 0.029) p<0.05; aircraft noise β=0.007 (95% CI 0.002, 0.017) p<0.05).
Fisher et al (2022)	Noise Pollution – Perceived Biodiversity – Mental Wellbeing Noise Pollution – Actual Biodiversity – Mental Wellbeing	No mediation observed. Although traffic noise pollution was associated with actual species richness (one of three objective measures of biodiversity), which was in turn associated with mental wellbeing, the Monte Carlo confidence interval for the indirect effect included zero (−0.134, 0.057)
Gomm & Bernauer (2023)	Traffic Noise – Perceived Noise Exposure – Mental Health	Evidence of an indirect effect of noise pollution via perceived noise exposure. Standardised regression coefficient β=0.026 (95% CI: 0.019, 0.033), p<0.001.
Klompmaker et al (2019)	Road Traffic Noise – Diabetes – Mental Health   Road Traffic Noise – Hypertension – Mental Health Road Traffic Noise – Asthma + COPD – Mental Health Road Traffic Noise – Stroke – Mental Health Road Traffic Noise – Heart Attack – Mental Health	No mediation observed.
Nazneen et al (2020)	Noise levels – Headaches – Anxiety	Noise exposure was associated with headaches (β=0.43) which was in turn associated with anxiety (β=0.59). CIs and P values not reported.
Oiamo et al (2015)	Environmental Noise – Noise Annoyance – HRQoL	Noise exposure was associated with noise annoyance (r=0.17, p<0.05) which was in turn associated with mental health (r=-0.17, p<0.05).
Peltz & Rogge (2016)	Noise Disturbances – Sleep Disturbance – Depressive Symptoms	Among university students, environmental noise was associated with sleep disturbance (β=0.20 if living alone, β=0.22 if living with roommates), which predicted higher depressive symptoms (β=0.27 living alone, β=0.20 living with roommates; all p<0.05). Indirect effect estimates were statistically significant (living along: 0.05 (95% CI: 0.01, 0.10); with roommates: 0.04 (95% CI: 0.01, 0.09)).
Peltz (2020)	Night-time Noise Disturbances – Restedness – Depressive/Anxiety Symptoms	Indirect effect of environmental noise disturbances on students’ depressive/anxiety symptoms was significant: the presence of nighttime noise disturbances predicted lower levels of restedness, which in turn predicted depressive/anxiety symptoms (between-student indirect effect=0.23, 95% CI: 0.07,0.38).
Roswall et al (2015)	Traffic Noise – Waist circumference + Smoking Status + Alcohol Intake – Mental Health	While not tested with formal mediation analysis techniques, the authors concluded there was possible mediation by one or more of waist circumference, smoking status and alcohol intake (analysed only as a block), based on the association between traffic noise and mental health being weakly attenuated, from -0.24 (95% CI: -0.37,-0.11) to -0.18 (95% CI: -0.31,-0.05), after adjustment for these lifestyle factors.
Stansfeld et al (2021)	Traffic Noise – Noise Annoyance – Psychological Ill-health	No evidence that noise annoyance was a mediator of the association between environmental noise and mental health.

Supplementary Table S4.

Summary of characteristics of included studies

Study characteristic	Number of Studies	Percentage
Location   Europe   North America   Asia	10 3 1	71.4% 21.4% 7.2%
Study Design   Cohort    Cross-sectional	4 10	28.6% 71.4%
Exposure    Transport Noise    Non-transport noise	11 3	78.6% 21.4%
Potential mediators examined    Noise annoyance    Sleep disturbance/quality/restedness    Physical activity    Neighbourhood restorative quality    Neighbourhood social cohesion    Perceived noise pollution    Biodiversity (actual/perceived)    Body mass index    Chronic physical disease    Headaches    Waist circumference    Smoking status    Alcohol intake	9 4 1 1 1 1 1 1 1 1 1 1 1	64.3% 28.6% 14.4% 7.2% 7.2% 7.2% 7.2% 7.2% 7.2% 7.2% 7.2% 7.2% 7.2%
Sample Size   < 500    500 – 999    1000 – 1999    2000 – 2999    3000 – 3999    4000 – 4999    5000 – 5999    6000 +	4 3 1 2 0 1 1 2	28.6% 21.4% 7.2% 14.4% 0% 7.2% 7.2% 14.4%

Across the 14 studies, 13 distinct mediating factors were analyzed. Some studies examined multiple mediators. The mediators were noise annoyance (eight studies), sleep (four studies, measured as a disturbance (three) and/or restedness (one), neighborhood restorative quality (one study), neighborhood social cohesion (one study), physical activity (two studies), body mass index (BMI) (one study), waist circumference (one study), smoking status (one study), alcohol intake (one study), biodiversity (perceived and actual, one study), perceived noise exposure (one study), chronic physical disease (one study), and headaches (one study). The number of mediators included in each study varied from one to six.

The most common method of measuring residential noise levels was using address data to calculate the distance from a person’s place of residence to various noise sources such as roads, railways, and airports (n = 10). Specific calculations used to determine noise levels varied. Other studies used noise sensors mounted at or near the home to measure noise levels (n = 2) or self-reported measures of noise (n = 2).

While most studies focussed on noise from transport and industrial sources, two studies asked participants about noise coming from within the home itself.[27,32] Both studies were conducted in residential colleges, including students who lived with roommates.

Eight of the 14 studies were conducted on the general population. A further four studies specifically looked at university students. One study was undertaken in a cancer-free cohort of adults aged 50–64 years, and the remaining study included only males aged 45–59 years. Most studies measured self-reported mental health using validated mental health assessment surveys that allow for an accurate estimate of the participants’ level of mental ill-health.[35]

Mediating Factors

Noise Annoyance

Noise annoyance can most generally be defined as the individual’s emotional response to unwelcome noise.[36] This was by far the most studied mediator (n = 8).[11,12,23,24,27,29,32,34] Of these, one study[34] considered “perceived noise” exposure as a mediator, which was measured by asking participants to what extent they felt “burdened” or “not burdened.” We determined this was a measure of noise annoyance as it captures a negative emotional or attitudinal response to noise exposure. The other seven studies explicitly referred to noise annoyance [Tables 2 and S5].

Table 2.

Environmental Noise–Mental Health Studies Examining Mediation by Noise Annoyance.

Study	Exposure	Outcome	Mediation Observed?
Bakker et al.[23]	Wind turbine noise	Psychological distress	Yes
Cerletti et al.[29]	Aircraft noise Road noise Train noise	Mental health	No
Dzhambov et al.[12]	Residential noise	Mental ill-health	Yes, in sequential mediation pathways
Dzhambov et al.[27]	Traffic noise	DepressionAnxiety	Yes, in individuals with low noise sensitivity (possible effect modification)
Eze et al.[11]	Aircraft noiseRoad noiseTrain noise	Depression	Yes, for aircraft and road traffic noise
Gomm and Bernauer[34]	Traffic noise	General mental health	Yes
Oiamo et al.[24]	Environmental noise	Mental health	Yes
Stansfeld et al.[32]	Traffic noise	Psychological ill-health	No

Of these eight studies, two were conducted on undergraduate students.[12,27] Five used a cross-sectional design, and three were cohort studies. Sample sizes ranged from 437 to 5729, with seven studies including both males and females and one study[32] including only male participants aged 45–59 years. All but one study focussed on traffic noise as the exposure, with the single exception focussing on noise from wind turbines.[23] All the studies estimated noise levels at each participant’s residential address. Data on noise annoyance was consistently measured using short questionnaires of either one or two questions, wherein participants were typically asked how often they were annoyed by noise or to rate how annoying they found the noise levels at their place of residence. None of the questionnaires separated annoyance in response to daytime and nighttime noise, though one asked about annoyance in response to noise experienced outside and inside the place of residence separately.[23] Three of these eight studies also examined noise annoyance either in sequence or parallel with other mediators.[11,12,23]

Of the eight studies that looked at noise annoyance as a single mediator, five found that noise annoyance independently mediated the association between environmental noise and mental health.[11,23,24,27,34] In one case, noise annoyance mediated the mental health impact of road traffic and aircraft noise but not of railway noise.[11] Additionally, while Dzhambov et al.[12] (in their 2018 study) observed no indirect pathway solely through noise annoyance; in sequential mediation analyses with sleep and separately with restorative quality and physical activity together, noise annoyance was found to be a significant mediator. In comparison, Cerletti et al.[29] and Stansfeld et al.[32] found no evidence of noise annoyance mediating the association between environmental noise and mental health. Notably, the population in the latter study was limited to males between 45 and 59 years, while the other studies looked at all adults.

In their 2019 study, Dzhambov et al.[27] examined noise sensitivity as a moderator of the mediation pathway through noise annoyance and found noise annoyance to be a more important mediator in people with low noise sensitivity. Eze et al.[11] found that in a parallel mediation model including noise annoyance, BMI, and physical activity, noise annoyance was the only significant mediator of the mental health impact of road traffic noise and aircraft noise, but it did not mediate any mental health effects of railway noise. Taken together, we conclude that noise annoyance may be a mediating factor in the association between noise pollution and mental health. However, its importance may not be the same across all populations or noise sources, and other factors, such as noise sensitivity, may be relevant moderators.

Sleep

Three studies investigated sleep disturbance as a potential mediator,[12,23,26] and another study investigated mediation by “restedness”[31] [Tables 3 and S5]. Sleep disturbance was self-reported, with one study[26] using the Pittsburgh Sleep Quality Index tool and the other two using one or two standalone survey questions.[12,23] Restedness was measured through a single-question survey completed by participants in the morning.[31] Three of the four studies were conducted in university student populations, with one examining transport noise and the other two measuring self-reported noise within residential colleges, including noise from within dormitories, shared apartments, and corridors.[12,26,31] In the fourth study, Bakker et al.[23] investigated wind turbine noise in members of the general community. Two of the sleep mediation studies[26,31] used self-reported noise as the exposure, while the other two used address data to determine distance from noise sources and the resulting noise exposure.

Table 3.

Environmental Noise–Mental Health Studies Examining Mediation by Sleep.

Study	Exposure	Outcome	Mediation Observed?
Bakker et al.[23]	Wind turbine noise	Psychological distress	No
Dzhambov et al. [12]	Residential noise	Mental ill-health	Yes, in sequential mediation following noise annoyance
Peltz and Rogge[26]	Self-reported noise disturbances	Depressive symptoms	Yes
Peltz.[31]	Self-reported night-time noise disturbance	Depressive/Anxiety symptoms	Yes

Sample sizes of the four studies ranged from 257 to 725, with all four studies including both male and female participants. All used a cross-sectional design and structural equation modeling for the analysis. Peltz and Rogge[26] and Peltz[31] looked exclusively at aspects of sleep, while the other studies looked at sleep disturbance amongst other mediating factors. Dzhambov et al.[12] only reported a sequential mediation model whilst Bakker et al.[23] reported single and parallel mediation models.

Two of the studies examined sleep disturbance in a single mediator model.[23,26] Bakker et al.[23] investigated sleep disturbance in those who reported that they did not notice the noise of nearby wind turbines. Peltz and Rogge[26] looked at sleep disturbance as a mediator in both students with and without roommates in the residential college. Whilst Bakker et al.[23] determined that sleep disturbance was not a mediator of the association between environmental noise and mental health, Peltz and Rogge[26] concluded that it was. Consistent with Peltz and Rogge,[26] Peltz[31] showed that restedness also played a significant mediating role in the association. Given the difference in study populations (general adult population vs. university students in residential colleges) and noise sources (wind turbine vs. dorm room) and measurement (geographically determined vs. self-reported), direct comparisons are not possible, and the apparently contradictory results may indicate that sleep is only an important mediator in some contexts.

Dzhambov et al.[12] looked only at the mediation effect of sleep disturbance in sequence with noise annoyance and found that sleep disturbance arising from noise annoyance significantly mediated the association between environmental noise and mental health. Bakker et al.[23], in contrast, concluded that while sleep disturbance did arise from noise annoyance, sleep was not a significant mediator of the effect of wind turbine noise on psychological distress.

Overall, it appears that sleep disturbance is likely linked to noise annoyance and may act as a mediator between noise exposure and mental health (such as in university residential colleges); however, the cross-sectional design of all relevant studies is an important limitation as these studies could not ensure temporal precedence of the exposure over the mediator and the mediator over the outcome (i.e., that noise exposure preceded any change in sleep and change in sleep preceded any change mental health). Evidence from a single study of wind turbine noise suggests sleep disturbance may not mediate any relationship between wind turbine noise and mental health.[23]

Physical Activity

Two studies investigated whether the reduction in physical activity levels mediates the association between environmental noise exposure and mental health [Tables 4 and S5].[11,12] The studies were both cross-sectional designs (n = 720 and n = 4581) and included male and female participants. Both studies focussed on transportation noise as the exposure of interest, measured using address information, and multiple mediators were explored in both studies.

Table 4.

Environmental Noise–Mental Health Studies Examining Mediation by Physical Activity.

Study	Exposure	Outcome	Mediation Observed?
Dzhambov et al.[12]	Residential noise	Mental ill-health	Yes, in sequential mediation with both noise annoyance and neighborhood restorative quality
Eze et al.[11]	Aircraft NoiseRoad NoiseTrain Noise	Depression	Yes, in sequential mediation with both noise annoyance and BMI (road traffic only)

Note: BMI, body mass index.

Eze et al.[11] looked at physical activity in both a single-mediator model and a multimediator model with both parallel and sequential indirect effects. They found that physical activity by itself was not a mediator of the association between noise exposure and mental health, regardless of whether the main source of the noise was road, railway, or air traffic. Dzhambov et al.[12] found physical activity to have a sequential indirect effect when combined with noise annoyance and neighborhood restorative quality but not combined with noise annoyance and sleep. Physical activity was also found by Eze et al.[11] to have a sequential indirect effect when combined with noise annoyance and BMI when the source of noise exposure was road traffic; however, in this model, the only mediating factor to have a significant effect individually within the parallel model was noise annoyance.

Overall, it is not possible to draw strong conclusions from the two studies, especially given Dzhambov et al.[12] only included medical students in their sample, reducing the generalisability of the results. However, it does not appear likely that physical activity by itself plays an important mediating role in the association between environmental noise exposure and mental health. It may contribute some effect downstream of other potential mediators. Further research would be required to achieve a more definitive understanding of physical activity’s role (or lack thereof) in explaining the association between environmental noise exposure at home and mental health.

Other Mediators

None of the remaining 10 proposed mediators were included in more than one study, and many were only included in models of parallel mediation or sequential mediation. The studies reported evidence of neighborhood restorative quality[12] and headaches[30] acting as mediating factors of environmental noise exposure and mental health outcomes [Tables 5 and S5]. Neighborhood restorative quality was only studied in sequential mediation with noise annoyance, so whether it also acts independently of noise annoyance remains unknown.[12] Similarly, BMI was only studied in sequential mediation with noise annoyance and physical activity; it was not found to be a significant factor in this pathway.[11] Although Fisher et al.[33] found traffic noise pollution was associated with plant biodiversity, which was, in turn, associated with mental well-being; in mediation analysis, the confidence interval for the indirect effect included zero.

Table 5.

Environmental Noise–Mental Health Studies Examining Mediation by Other Factors.

Mediator	Study	Exposure	Outcome	Mediation Observed?
Neighborhood restorative quality	Dzhambov et al.[12]	Residential noise	Mental ill-health	Yes, in sequential mediation with noise annoyance
Neighborhood social cohesion	Dzhambov et al.[12]	Residential noise	Mental ill-health	No
Body mass index	Eze et al.[11]	Aircraft noiseRoad noiseTrain noise	Depression	No
Actual biodiversity	Fisher et al.[33]	Road traffic noise	Mental wellbeing	No
Perceived biodiversity	Fisher et al.[33]	Road traffic noise	Mental wellbeing	No
Chronic physical illness	Klompmaker et al.[28]	Road traffic noise	Mental ill-health	No
Headaches	Nazneen et al.[30]	Noise levels	Psychological symptoms	Yes
Waist circumference, smoking status, alcohol intake	Roswall et al.[25]	Traffic noise	Mental health; HRQoL	Possible mediation by one or more of waist circumference, smoking status, and alcohol intake (analyzed only as a block)

Note: HRQoL, health-related quality of life.

A study by Roswall et al.[25] included waist circumference, smoking status, and alcohol intake as covariates in the models, nominally as potential confounders; however, upon finding they did attenuate the association between road traffic noise and mental health, the authors speculated that these factors may partially mediate relationships between road traffic noise exposure and mental health, rather than (perhaps less plausibly) confound them. The attenuation of the exposure effect estimate after adjustment for a potential mediator is consistent with one traditional approach to identifying mediation (proposed by Baron and Kenny[37] and widely used in epidemiology until recently). This distinction could not be disentangled given the analysis undertaken and the uncertain temporal sequencing of relevant variables, and the authors only report models where all three variables were modeled as a block, precluding any comment about individual mediating roles. No other proposed mediator was found to be part of a significant indirect pathway in any study.[12,28,33] Many of the studies lacked a stated rationale for the proposed mediators.

Risk of Bias

The risk of bias was a concern for all studies [Table S6]. All studies were assessed as having at least some risk of both selection bias and information bias due to possible measurement error in the exposure or outcome. Four studies also had a moderate or high risk of bias due to failure to control for one or more important confounders[23,26,30,31]. In all studies, there was also some nonnegligible risk of bias arising from the mediation component of the analysis, resulting from one or more of failures to account for exposure-mediator interaction, potential measurement error in the putative mediator, or inadequate consideration of mediation-related causal assumptions.

Supplementary Table S6.

Results of risk of bias analysis

		Selection Bias		Information Bias			Confounding	Mediation-related
Study	Study design	Risk of selection bias (in design or analysis)	Response bias or loss to follow-up	Exposure: Reliance on area-level data	Exposure: risk of measurement error	Outcome: risk of measurement error	Outcome: Standardised/ validated tool not used	Failure to adjust for important confounder(s)	Risk of measurement error in mediator	Lacks graphical representation of hypothesised mediation pathways	Exposuremediator interaction not accounted for or discussed	Lacks description of causal assumptions of mediation analysis
Mediator: Noise annoyance
Bakker 2012	Crosssectional		X			X		X	X		X	X
Cerletti 2020	Cohort	X	X			X			X		X	X
Dzhambov 2018	Crosssectional	X	X		X	X			X		X	X
Dzhambov 2019	Crosssectional	X	X		X	X			X	X	X	X
Eze 2020	Cohort	X	X			X			X		X	X
Oiamo 2015	Crosssectional		X			X			X		X	X
Stansfeld 2021	Cohort	X	X		X	X			X	X	X	X
Mediator: Sleep
Bakker 2012	Crosssectional		X			X		X	X		X	X
Dzhambov 2018	Crosssectional	X	X		X	X			X		X	X
Peltz & Rogge 2016	Crosssectional	X	X		X	X		X	X		X	X
Peltz 2020	Crosssectional	X			X	X	X	X	X		X	X
Mediator: Physical Activity
Dzhambov 2018	Crosssectional	X	X		X	X			X		X	X
Eze 2020	Crosssectional	X	X			X			X		X	X
Mediator: Other
Dzhambov 2018 (n’hood restorative quality )	Crosssectional	X	X		X	X			X		X	X
Dzhambov 2018 (n’hood social cohesion)	Crosssectional	X	X		X	X			X		X	X
Eze 2020 (BMI)	Crosssectional	X	X			X					X	X
Fisher 2022 (actual biodiversity)	Crosssectional		X			X					X	X
Fisher 2022 (perceived biodiversity)	Crosssectional		X			X			X		X	X
Gomm 2023 (perceived noise pollution)	Crosssectional		X			X			X		X	X
Klompmaker 2019 (chronic physical illness)	Crosssectional	X	X			X			X	X	X
Nazneen 2020 (headaches)	Crosssectional	X	X			X	X	X	X		X	X
Roswall 2015 (waist circumference, smoking, alcohol)	Cohort	X	X			X			X		X	X

Certainty of Evidence

The certainty of the evidence was assessed using the GRADE assessment tool. The observational nature of the studies (most of them using a cross-sectional design) and the medium-high risk of bias across most studies resulted in no criteria for upgrading being met [Table 6]. As such, the evidence was of low or very low certainty across all mediators, indicating that caution should be taken when interpreting the results.

Table 6.

Summary of GRADE Analysis of Studies Included in the Systematic Review.

Potential Mediator	Mediation Effect	Number of Participants (Studies)	Certainty in the Evidence
Noise annoyance	Most studies report evidence in support of mediation by noise annoyance	10,421 (9)	Low (downgraded primarily due to risk of bias)
Sleep	Studies of university students report evidence in support of mediation by sleep	2037 (4)	Very low (downgraded due to indirectness, risk of bias)
Physical activity	Evidence only of mediation in sequence with noise annoyance	5301 (2)	Very low (downgraded due to imprecision, indirectness, risk of bias)
Other mediators	Limited evidence	N/A (all single studies)	Very low (downgraded due to imprecision, indirectness, risk of bias)

GRADE = Grading of Recommendation, Assessment, Development and Evaluation.

DISCUSSION

Key Findings

This review aimed to identify factors that mediate the association between environmental noise exposure at home (or in other residential settings such as university colleges) and mental health. Analysis of 14 studies encompassing 13 unique mediating factors revealed a limited understanding of the pathways between environmental noise exposure and mental health. With each mediating factor only included in between one and nine studies and a high risk of bias in most studies, few firm conclusions about mediating pathways can be drawn from our review.

Our synthesis indicates that noise annoyance and sleep disturbance are likely to play a role in mediating the association between noise exposure at home and mental health outcomes. Evidence for sleep disturbance as a mediator is limited to student populations; whether findings are generalizable to the wider adult population is unknown. Physical activity may act as a mediator when combined with other mediators. Albeit on the basis of a very small number of studies, none of the other proposed mediators examined in the literature were shown empirically to be significant. Additionally, a strong rationale was lacking for many of these other variables being plausible mediators.

Noise annoyance

Studies over the last decade have shown that noise annoyance is significantly associated with negative mental health outcomes, and for many noise sources, there is a clear dose–response relationship between noise exposure and noise annoyance.[38] Animal and human studies of biological mechanisms linking noise and mental health indicate that a mediating pathway through noise annoyance very likely involves hyperactivation of the hypothalamic–pituitary–adrenal axis and sympathetic nervous system, leading to a cascading stress response resulting in systemic inflammation and neuroinflammation. Prolonged exposure to these stress responses is a known risk factor for depression and anxiety.

Evidence from our review largely supports noise annoyance as a mediator of the association between environmental noise at home and mental health. However, the mediating role of noise annoyance appears to be dependent on the source of noise and may not mediate the mental health impacts of railway noise.[11] This aligns with other European based studies that find that overall railway noise is considered less annoying than road or air traffic noise.[39] This, however, has not been found to be the case in Japan, where railway noise has been found to be at least as annoying as road or air traffic noise.[40] If railway noise does elicit less annoyance, this may be attributable to either a physical characteristic of the noise, such as regularity in train timetables and the predictable noise level, or the extended periods without noise in between each train passing, unlike the noise of cars on a busy road. An alternative explanation is the more positive societal attitude towards railways and trains, as in many cultures, people feel sentimental about trains or approve of them due to their convenience and environmental contribution; for example, someone may choose to live near a train line (and tolerate or mitigate the associated noise) so that they can conveniently commute to work—something which cannot generally be said for someone living near an airport runway or busy freeway.

The cross-sectional design of many of the studies in this review rendered them unable to tease apart the possible bi-directional relationships between noise annoyance and mental health. It has been suggested that having existing poor mental health could lead to greater noise annoyance, just as noise annoyance might contribute to a subsequent mental health outcome.[41] Longitudinal studies are needed to unpack this potentially complex relationship and better understand the role of noise annoyance in relation to environmental noise and mental health.

Sleep

Based on the studies identified in our review, there is no sufficient evidence to determine whether either sleep disturbance or restedness mediates the association between environmental noise exposure and mental health outcomes in the general population. The studies available were limited to specific subsets of the population, and a variety of noise sources were explored across the studies. It appeared that in university college residents, sleep disturbance did mediate the association between self-reported noise from within the residence and mental health, although reverse causation cannot be excluded. On the other hand, Bakker et al.[23] found that in adults exposed to noise from wind turbines, sleep disturbance was not a significant mediator of the association between noise exposure and mental health; although they identified a significant positive association between noise annoyance and sleep disturbance, there was no direct association between noise exposure and sleep disturbance nor was there a significant association between sleep disturbance and mental health. Whether this finding is particular to wind turbine noise is unknown.

As with noise annoyance, mediation of the noise-mental health relationship via sleep disturbance very likely involves pathophysiological pathways involving the stress response.[10,42] In the case of sleep, dysregulation of the circadian clock is also thought to be involved, with clinical and animal studies lending support to such an explanation.[43] Future research should prioritize testing sleep-related mediators across diverse populations and noise sources using objective measures of both noise exposure and sleep disturbance to improve understanding of these pathways. Longitudinal studies are needed to determine whether sleep disturbance precedes mental health changes or arises as a consequence of pre-existing mental health issues (which may separately be causally related to noise annoyance).

Physical Activity and Other Proposed Mediators

Other studies have shown that behavioral risk factors for mental health disorders, such as physical inactivity, smoking, and alcohol consumption, are associated with traffic noise. It has been proposed that these behaviors may represent maladaptive coping responses induced by chronic stress arising from noise annoyance and/or sleep disturbance.[42,44] We found two studies that investigated physical activity as a mediator of the association between noise exposure at home and mental health outcomes. While neither found that physical activity independently acted as a mediator, the findings suggest it may play a role in sequential mediation following noise annoyance. This would be consistent with a stress pathway mechanism affecting physical activity behaviors and with previous studies that found that reduced physical activity is more commonly associated with noise annoyance than the presence or absence of noise itself.[11,12] Future studies should focus on confirming or refuting this sequential mediating pathway.

Ten other potential mediators were identified; however, none of these were included in more than one study, making comparison with other studies impossible. Two of these mediators—headaches and plant diversity—were identified as significant mediators in their respective studies;[30,33] however, the rationale for these studies were insufficiently explicated, and one study used an unvalidated mental health measure,[30] so we argue these conclusions should be interpreted with caution. Dzhambov et al.[12] suggested a complex mediating path, including neighborhood restorative quality in conjunction with noise annoyance and physical activity in students. However, while there is an association between noise annoyance and neighborhood restorative quality, the direction of effect is not clear. Further exploration of such complex, multistep pathways is warranted, but researchers should ensure that each step of the pathway is theoretically justified. Longitudinal study designs are essential for determining the temporal ordering of key variables. None of the remaining mediators tested in the sample of studies were found to be significant, and many studies lacked a clear rationale for the mediation pathways under consideration. To strengthen the evidence base, future studies should focus on identifying theoretically grounded mediators and apply causal methods to test their roles in the noise-mental health relationship.

Limitations of the Evidence Base and Implications for Future Research

Our certainty in the evidence overall is low, driven by a reliance on observational study designs (many with small samples and/or narrowly defined study populations) and the risk of bias from various sources. A meta-analysis was considered infeasible in this review due to the variety of mediators included and the fact that many mediators were only included in one study. Additionally, where multiple studies existed for a single mediator, there was a range of different analysis methods across the studies, such as structural equation modeling (SEM) and traditional mediation methods, that prevented aggregation of results. This heterogeneity across studies also limits our ability to draw generalizable conclusions. There is also a risk of publication bias, although this is difficult to assess with so few studies for most mediators. We found only one eligible study conducted outside of Europe or North America. We included only articles available in English due to limited resources for translation, and it is possible this led to the exclusion of studies set in non-English speaking countries and/or published in non-English journals. Contextual factors—such as local norms and predominant noise sources—could plausibly drive effect heterogeneity with regard to the mental health effects of environmental noise. It is, therefore, important that the association between noise pollution and mental health outcomes and its mediating pathways be further investigated across a variety of geographical and cultural contexts.

The majority of studies were cross-sectional in design and, therefore, unable to address issues of reverse causation and establish the appropriate temporal sequencing of exposure, mediator, and outcome. Most studies were also at risk of selection bias due to nonresponse, attrition, or sampling methods. In addition, and specific to the analysis of potential mediation, very few of the included studies gave sufficient consideration to mediation-related causal assumptions, either through choice of analytic method or explicit assumption recognition or testing.

Information bias linked to exposure and/or mediator measurement is a risk in most of the included studies. Many estimated noise levels at home by measuring the distance between the participant’s home and the noise source. However, this approach may not accurately capture variation in true noise exposure due to residence-level variables such as insulation, home layout, and internal noise sources (e.g., appliances and household members). Studies relying on self-reported measurement of noise exposure, noise annoyance, sleep disturbance, and other potential mediators faced the additional risk of recall bias. Few studies considered the amount of time spent at home and exposure to additional noise sources outside the home, such as occupational or leisure noise, which may also have influenced the outcome.

As most of the studies in this review investigated the impact of transport noise, it is crucial to recognize that this often coexists with air pollution, dense urbanization, and other environmental factors, raising the possibility of confounding by these variables. Some of the studies controlled for air pollution as a potential confounder; however, for many, there remains a nonnegligible risk of unmeasured confounding by other strongly correlated factors, and this impinged on the reliability of the associated mediation analyses.

With such a small number of studies available for each mediator and the overall high risk of bias across these, results from this review should be interpreted with caution, especially in more diverse populations who were not represented. There is a clear need for more high-quality studies on this topic in order to adequately assess the pathways through which noise impacts mental health, in order to better guide the development of effective interventions to address this important public health concern. In particular, future studies should ideally employ longitudinal designs (e.g., high-quality cohort studies, quasi-experimental methods) applied to the general population and take advantage of natural experiments in order to better establish of the temporal sequencing of the exposure, mediators, and outcomes and thereby strengthen the ability to draw causal inferences. Similarly, the use of causal mediation methods and greater attention to the causal assumptions underlying traditional methods of analyzing mediation would substantially improve the evidence base on this topic. Future research should also focus on geographical settings not currently well represented in the evidence base and, wherever possible, use objective or validated measures of both noise exposure and mediators, for example, measuring noise exposure using environmental sensors within the home and physiological and behavioral mediators such as sleep using accelerometers and other wearable devices.

Finally, with noise annoyance and sleep emerging as the most prominent and plausible mediators of the noise-mental health relationship, future studies should investigate what characteristics or noise sources cause greater annoyance and sleep disturbance across diverse populations and whether there are cross-cultural differences. Multiple mediations (parallel or in sequence) by sleep disturbance and noise annoyance should also be further examined to determine their individual and joint mediating pathways and relative importance.

CONCLUSIONS

Noise annoyance, sleep, and physical activity are the most frequently studied potential mediators of the association between environmental noise and mental health. However, research on these mediating relationships remains limited, and study quality varies. While some evidence suggests that noise annoyance and sleep independently mediate the noise-mental health relationship, these results should be interpreted with caution due to the low certainty of evidence. Many studies in this review used cross-sectional designs and were limited by either a lack of generalisability or likely exposure measurement error. Additionally, few studies gave appropriate consideration to mediation-related causal assumptions, and the synthesis of results was hampered by heterogeneity in the measurement of noise exposure and mediating factors.

To establish more robust conclusions about the clinical and public health significance of noise annoyance and sleep disturbance as mediators of the mental health effects of noise exposure, high-quality evidence from longitudinal and quasi-experimental studies is needed. Further research is also needed to determine whether other important mediators exist. A deeper understanding of the pathways linking environmental noise exposure to mental health outcomes will assist in developing and targeting effective interventions.

Protocol registration

Our review protocol was registered in the International Prospective Register of Systematic Reviews (PROSPERO) with a register number CRD42023416773.

Author Contributions

Kaya Grocott (guarantor): conceptualization, study design, literature search, data curation, analysis, manuscript preparation and editing. Adelle Mansour: manuscript review and editing, supervision. Eleanor Shiels: literature search, data curation, manuscript review. Rebecca Bentley: Manuscript review and editing, supervision. Kate Mason (guarantor): conceptualization, study design, literature search and data curation, manuscript writing, review and editing, supervision.

Availability of Data and Materials

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Conflicts of Interest

The authors declare they have no conflicts of interest related to this work to disclose.

Funding Statement

We acknowledge funding from the National Health and Medical Research Council (NHMRC Centre of Research Excellence in Healthy Housing APP1196456).