The relationship between ambient neighborhood noise exposure and sleep parameters among Black adults living in the Miami metropolitan area

Abstract

Objective:

Black individuals in the US are burdened by sleep health disparities compared to non-Hispanic White individuals. Neighborhood factors could contribute and provide insights into these disparities. This investigation examined associations between environmental noise and sleep parameters among Black adults to deepen the understanding of sleep disparities.

Methods:

Participant sleep parameters were recorded at-home for seven nights using cardiopulmonary coupling-based SleepImage Ring. Soundscores^™ derived from HowLoud were used as a proxy for neighborhood noise. Score are noise estimates based on traffic models and local factors. Participants on average lived in slightly noisy neighborhoods (soundscore M = 70.8). Linear regressions were conducted to assess the relationship between noise and sleep parameters.

Results:

Analysis of 261 Black adults (aged 18–85 years; 66.7 % female) from South Florida showed quieter environments (higher soundscores) were associated with increased deep sleep and improved sleep quality. When stratified by age and sex, lower noise was associated with decreased nocturnal awakening duration in males aged 18–54 years (β [95 % CI] = −110.76 [−203.18, −18.34]; p = .02; r² = .20) and increased time slept in females aged 55–85 years (β [95 % CI] = 379.48 [5.92, 753.04]; p = .04; r² = .09).

Conclusions:

To the authors’ knowledge, this investigation is among the first to investigate associations between noise and sleep in Black adults using objective measures. Results showed sleep parameters were associated with noise, with unique demographic variations. Future investigations are necessary to address the effects of neighborhood features on sleep among Black adults.

1. Introduction

Racial disparities in sleep health in the United States are well-documented, with Black adults and other individuals from minoritized groups often experiencing poorer sleep outcomes compared to non-Hispanic White adults [1]. Traditionally, these disparities have been assessed primarily through sleep duration, with Black individuals frequently reporting shorter sleep relative to other racial/ethnic groups [2]. This has been associated with downstream adverse health outcomes such as heightened risks for cardiovascular and metabolic diseases [3]. Yet, the underlying causes and antecedents of these disparities in specific sleep parameters remain unknown [4]. While socioeconomic and psychosocial stressors, such as financial insecurity, discrimination, and neighborhood disadvantage have been suggested as contributing factors, these influences only partially explain the observed differences [5]. Thus, there is a need to explore additional factors that may account for these gaps, particularly environmental influences that impact several sleep parameters as few studies have investigated this relationship in this population [6]. More explorations into the environmental determinants of sleep in Black adults are needed to understand better and address sleep health disparities.

One critical but understudied environmental factor is noise pollution, which is becoming more important with continued global urbanization [7]. Black individuals in the US are more likely to reside in areas that are disproportionately exposed to higher levels of environmental noise [8]. This is often attributable to historical and systemic factors that have concentrated Black populations in urban areas where noise levels from a variety of sources tend to be higher [9]. Exposure to noise while asleep has been shown to disrupt sleep by increasing sleep fragmentation, reducing time in REM (rapid eye movement) and deep sleep (N3), and ultimately decreasing overall sleep efficiency [10].

Despite the many insights, the existing body of research on the relationship between sound and sleep is limited in that it draws mostly from Northern European study populations [11]. This is problematic given that previous work has shown that noise-induced sleep disturbances vary among individuals of different racial/ethnic groups [12]. Given the established connection between environmental noise and sleep disruption, it is plausible that noise pollution could play a significant role in driving sleep disparities among Black individuals. Despite this potential link, existing studies have not examined the relationship between environmental noise and sleep parameters among Black adults using objective measures [13]. Addressing this gap could provide a crucial understanding of how environmental factors contribute to racial disparities in sleep health. This study aims to investigate how neighborhood noise affects sleep parameters in Black adults, offering insights into an often-overlooked dimension of sleep inequality.

2. Methods

2.1. Participant and procedures

2.1.1. Participants

The data used in the study were obtained from two NIH-funded studies: Determinants of Insufficient Sleep Among Blacks and Effects on Disparities in Health Outcomes (ESSENTIAL) and Mechanisms of Sleep Deficiency and Effects on Brain Injury and Neurocognitive Functions Among Older Blacks (MOSAIC). Both studies were conducted in accordance with the Declaration of Helsinki. The respective Institutional Review Boards for New York University and the University of Miami granted each study’s ethical approval. Informed consent was obtained from all participants. Participants self-identified as Black, including African American, Caribbean American, and other African backgrounds. Participants lived in either the New York City or Miami metropolitan area at the time of the data collection. In this present study, the analysis was solely focused on Miami participants as investigating this topic among New York City residents has previously been conducted by other researchers (see citation 14) [14]. Of note, the age requirements for enrollment differed for each study; ESSENTIAL participants were adults aged ≥ 18 years, and MOSAIC participants were restricted to adults aged 55–85 years. Outside of age, the other requirements were that participants had no existing history of sleep problems and were deemed cognitively fit enough to partake in and complete the study. The latter was evaluated through the Callahan six-item screener [15].

2.1.2. Procedures

While ESSENTIAL and MOSAIC are on-going, data utilized for this investigation were collected between September 2021 and October 2024. The studies were done in phases. During the baseline phase, participants were administered surveys and questionnaires using the Research Electronic Data Capture (REDCap) data management platform [16]. This collected information regarding demographics, socioeconomics, home environment, medical history. Following baseline, participants received a SleepImage Ring device and were instructed to wear it for seven nights in order to record and gather sleep data.

2.2. Measures

2.2.1. SleepImage ring

The wearable SleepImage Ring device captured participant sleep parameters. Wearable devices are typically favored over in-laboratory polysomnography (PSG) when investigating environmental factors because field surveys more accurately reflect an individual’s natural environment and sleep patterns [17]. The SleepImage Ring device utilizes photoplethysmography (PPG) to record users’ sleep. The device has a single-lead ECG sensor that simultaneously measures plethysmogram (PLETH) and SpO₂ data [18]. PPG technology captures PLETH signals by using light-based sensors to detect blood volume changes [19]. The SleepImage assess sleep parameters through cardiopulmonary coupling (CPC) calculations [20]. Sleep parameters derived using the SleepImage Ring PPG system shows high coherence with those determined via PSG [21]. The device provides sleep measures including sleep onset latency (SOL), wake after sleep onset (WASO), total sleep time (TST), sleep duration (SD), and sleep efficiency (SE) [22]. SOL is the time it initially takes to fall asleep, WASO (also referred to as nocturnal awakenings) is the time spent awake during a sleep period after initiation of sleep, and TST is the time during a sleep period actually spent asleep. SD is equivalent to the sum of TST and WASO, while SE is the ratio between TST and total time in bed (which is the sum of SD and SOL) and is represented as a percentage [23]. The system also measures sleep architecture based on CPC patterns that correlate to those determined via PSG. The three patterns are low frequency coupling (LFC), high frequency coupling (HFC), and very-low frequency coupling (vLFC). LFC is equivalent to N1 and parts of N2, HFC represents N3 and the remaining parts of N2, and vLFC equates to both REM sleep and wakefulness [24]. For this investigation, LFC and HFC will be referred to as light NREM and deep NREM, respectively. The Ring uses actigraphy to distinguish between REM and wakefulness [25]. The final parameter given is the sleep quality index (SQI). This a composite measure summarizing all sleep metrics ranging from 0 (low quality)–100 (high quality) [26].

2.2.2. Soundscore^™

Estimates of neighborhood noise levels were determined through HowLoud. This website calculates a Soundscore^™ (henceforth referred to as soundscore), which are estimations of the ambient noise level for a given address inputted. Soundscores are determined through a calculation that takes into account road traffic, air traffic, and other local factors present in the area assessed [27]. Road traffic values are based on the U.S. Federal Highway Administration Traffic Noise Model [28]. Local factors include things such as restaurants, schools, and stores [29]. Soundscores range from 50 (loud)–100 (quiet) [30].

2.2.3. Covariates

All covariates were based on participant self-report as collected and recorded from baseline surveys. Sociodemographic covariates include age (years), if they are employed or not, highest education level, and income. Income was based on a Likert scale from 1 to 6; in units of $1000: 〈 10, 10–19, 20–39, 40–59, 60–100, and 〉 100, respectively). Health-related covariates included were anti-depressant usage, anxiety diagnosis (dx), depression dx, and hypertension dx. For diagnosis, participants were asked if they have been diagnosed with the condition. The methodology used to determine covariate inclusion is detailed in the following section.

2.3. Analytical plan

2.3.1. Data processing

All data cleaning and analyses described in this section were performed in R Studio version 4.4.2 within the RStudio environment (version 2024.04.2 + 764) for MacOS. In order to maintain independence of observations, participant sleep data record across multiple nights was averaged so that each participant had one observation per parameter. This also ensured that participant data was equally accounted for in the results, regardless of the number of nights of sleep recorded (i.e., a participant not having a full seven nights of data). Prior to aggregation, data was first cleaned of problematic data points. In this case, records with abnormally low sleep durations, indicative of nap periods or removing the Ring while asleep, were excluded from data. These values were excluded to reduce the influence of extreme observations that may reflect measurement errors or atypical behaviors not representative of the study population. Sleep durations were excluded if they were <50 % of a participants’ average sleep duration.

2.3.2. Covariate selection

Demographic, health, and other potentially relevant variables were considered as covariates. Pearson correlations between variables and sleep parameters were conducted in order to determine which covariates would be controlled for in analyses. In order to maintain model parsimony, only covariates that were significantly correlated with a majority of (≥ five) the sleep parameters were selected. The significance threshold here (and for all statistical analyses not yet detailed) was set at α = 0.05. Covariate usage was kept consistent across models and subgroup analyses to preserve model homogeneity and facilitate more direct result comparison.

2.3.3. Statistical analysis

The relationship between soundscore and sleep parameters were assessed through linear regression analysis. Both simple and multiple linear regressions were conducted, with the latter adjusting for covariates. Additionally, analyses stratified by age group and biological sex assigned at birth were used. This was done because noise has been shown to differentially impact individuals depending on demographic characteristics. For example, women are often reported to be more sensitive to noise than men, which may lead to increased sleep disturbances [31]. Similarly, age-related hearing loss can alter noise perception and thus modify the relationship between noise and sleep [32]. Student’s t-tests were performed to explore the differences in sleep parameter (and covariates) among the demographic subgroups. Age groups classification was derived naturally from the age requirements of the two studies: aged 18–54 and aged 55–85 years. 55 years was kept as the age group cut-off because age-associated changes in hearing, which were previously discussed, typically present in the sixth decade of life [33]. Finally, all p-values obtained were corrected for multiple comparisons using the Benjamini-Hochberg procedure.

3. Results

This study analyzed sleep data from 261 participants. Each participant had on average 5.8 (SD = 2.2) nights of data. Table 1 summarizes participant covariate descriptive statistics. The average age for all participants in the sample was 49.7 (SD = 15.6) years (Table 1). While the participants were split almost evenly in terms of age group, the distribution of sexes was imbalanced, with approximately twice as many female participants as male participants. Overall, participants lived in moderately noisy neighborhoods, with an average soundscore of 70.8 (SD = 4.9) (Table 1). Soundscores did not significantly vary by age nor sex. Covariate characteristics among stratified groups are also presented in Tables S1–S4.

Table 1.

Study population characteristics.

Sleep parametera	M	SD	Continuous variable	M	SD
SOL (min)	19.9	25.9	Soundscoreb	70.8	4.9
WASO (min)	54.1	25.3	Age (years)c	49.7	15.6
TST (hrs)	5.7	1.5	Incomec,d	2.9	1.6
SD (hrs)	6.9	2.3	Nominal variable	n	%
SE ( %)	78.1	12.2	Employment statusc,e	110	42.2
Light NREM ( %)	39.5	16.6	Anxiety dxc,f	95	36.4
Deep NREM ( %)	39.3	20.8	Hypertension dxc,f	56	21.5
REM ( %)	21.2	7.2	Depression dxc,f	69	26.4
SQI	49.9	16.6	Anti-depressantsc,g	29	11.1

Note. N = 261 Black adults aged 18–85 years living in the Miami metropolitan area. SOL = sleep latency. WASO = nocturnal awakenings. TST = total sleep time. SD = sleep duration, (TST + WASO). SE = sleep efficiency, [TST / (SOL + SD)]. Frequency coupling = fc. REM = rapid eye movement (very-low fc). NREM = non-REM. Light NREM = low fc. Deep NREM = high fc. SQI =. sleep quality index (0–100).

^aMeasurements recorded and calculated via SleepImage Ring.

^bNoise estimate derived from HowLoud, ranges from 50 (quiet)–100 (loud).

^cUtilized as covariate in analyses. Participant self-reported.

^dLikert scale from 1 to 6. In units of $1000: ( 10, 10–19, 20–39, 40– 59, 60–100, and ) 100.

^eParticipant is employed.

^fParticipant is diagnosed with condition.

^gParticipant is taking medication.

This analysis examined sleep parameters across sex and age groups, combining data from all participants to identify overall trends. Overall study population sleep parameter averages are shown in Table 1. Notable differences were found among sleep parameters when comparing demographic groups. SD and REM were significantly different between participants aged 18–54 years and those aged 55–85 years (Table 2). Furthermore, REM was significantly different between the sexes among participants aged 18–54 years (Table 3). Differences in sleep parameters between the sexes among the whole study population and among participants aged 55–85 years are presented in Tables S5 and S6, respectively.

Table 2.

Sleep parameters among study participants stratified by age group.

Sleep parametera	Group 1b		Group 2c		Comparisond
	M	SD	M	SD	t	d	P
SOL (min)	17.0	21.3	23.0	29.7	−1.86	−0.23	.06
WASO (min)	52.1	22.3	56.2	28.1	−1.27	−0.16	.20
TST (hrs)	5.6	1.3	5.7	1.8	−0.57	−0.07	.57
SD (hrs)	6.6	1.5	7.3	2.9	−2.77	−0.35	.01*
SE ( %)	80.5	7.4	75.5	15.4	3.34	0.42	< .01*
Light NREM ( %)	34.4	16.7	44.8	14.7	−5.35	−0.66	< .01*
Deep NREM ( %)	45.4	21.6	33.0	17.9	5.05	0.62	< .01*
REM ( %)	20.2	7.4	22.1	6.8	−2.18	−0.27	.03*
SQI	55.1	17.4	44.4	13.8	5.54	0.68	< .01*

Note. N = 261 Black adults aged 18–85 years living in the Miami metropolitan area. SOL = sleep latency. WASO = nocturnal awakenings. TST = total sleep time. SD = sleep duration, (TST + WASO). SE = sleep efficiency, [TST / (SOL + SD)]. Frequency coupling = fc. REM = rapid eye movement (very-low fc). NREM = non-REM. Light NREM = low fc. Deep NREM = high fc. SQI = sleep quality index (0–100).

^aMeasurements recorded and calculated via SleepImage Ring.

^bParticipants aged 18–54 years, n = 133.

^cParticipants aged 55–85 years, n = 128.

^dDifferences in sleep parameters between age groups were assessed via student’s t-test.

^*p < .05, two-tailed.

Table 3.

Sleep parameters among study participants aged 18–54 stratified by biological sex.

Sleep parametera	Group 1b		Group 2c		Comparisond
	M	SD	M	SD	t	d	p
SOL (min)	15.3	11.1	20.8	34.9	−0.97	−0.26	.34
WASO (min)	52.9	21.7	50.3	24.0	0.60	0.12	.55
TST (hrs)	5.7	1.1	5.4	1.6	0.92	0.20	.36
SD (hrs)	6.7	1.4	6.3	1.8	1.17	0.24	.25
SE ( %)	80.9	6.5	79.6	9.3	0.82	0.18	.42
Light NREM ( %)	33.1	16.9	37.5	16.0	−1.46	−0.27	.15
Deep NREM ( %)	47.8	21.2	39.8	21.5	1.96	0.37	.05
REM ( %)	19.2	7.0	22.7	7.8	−2.48	−0.49	.02*
SQI	56.9	17.7	50.9	15.9	1.94	0.35	.06

Note. N = 133 Black adults aged 18–54 years living in the Miami metropolitan area. SOL = sleep latency. WASO = nocturnal awakenings. TST = total sleep time. SD = sleep duration, (TST + WASO). SE = sleep efficiency, [TST / (SOL + SD)]. Frequency coupling = fc. REM = rapid eye movement (very-low fc). NREM = non-REM. Light NREM = low fc. Deep NREM = high fc. SQI = sleep quality index (0–100).

^aMeasurements recorded and calculated via SleepImage Ring.

^bFemale participants, n = 93.

^cMale participants, n = 40.

^dDifferences in sleep parameters between sexes were assessed via student’s t-test.

^*p < .05, two-tailed.

Linear regressions revealed that environmental noise was significantly associated with aspects of sleep architecture and related parameters in Black adults. The results of these analyses on the full dataset are outlined in Table 4. Simple linear regressions among the full non-stratified dataset showed that soundscore was significantly associated with deep NREM, REM, and SQI. Deep NREM and SQI were observed to increase with increasing soundscore (i.e., the environment became quieter), while WASO decreased with increasing soundscore (i.e., the environment became louder). These associations were found to be non-significant after adjusted for covariates (Table 4).

Table 4.

Linear regression results for the association between soundscore and sleep parameters among all study participants.

Sleep parametera	β	SE	LLb	ULc	$R_{\text{adj}}^2$	p	pcd
Simplee
SOL (s)	−15.22	19.57	−53.75	23.31	< 0.01	.44	.87
WASO (s)	−34.31	19.06	−71.84	3.23	.01	.07	.37
TST (s)	−31.32	69.29	−167.76	105.12	< 0.01	.65	.87
SD (s)	−135.35	104.28	−340.70	69.99	< 0.01	.20	.59
SE ( %)	0.30	0.15	0.00	0.60	.01	.05	.35
Light NREM ( %)	−0.36	0.21	−0.77	0.04	.01	.08	.37
Deep NREM ( %)	0.58	0.26	0.07	1.10	.02	.02*	.20
REM ( %)	−0.21	0.09	−0.39	−0.03	.02	.02*	.17
SQI	0.41	0.21	0.00	0.82	.01	.05*	.35
Multiplef
SOL (s)	−9.14	20.87	−50.25	31.97	.05	.66	> 0.99
WASO (s)	−15.97	20.79	−56.91	24.97	.01	.44	> 0.99
TST (s)	−11.03	74.55	−157.84	135.78	.03	.88	> 0.99
SD (s)	−87.48	112.11	−308.27	133.31	.04	.44	> 0.99
SE ( %)	0.21	0.17	−0.12	0.53	.03	.21	> 0.99
Light NREM ( %)	−0.01	0.20	−0.41	0.40	.19	.98	> 0.99
Deep NREM ( %)	0.06	0.25	−0.44	0.55	.23	.81	> 0.99
REM ( %)	−0.04	0.09	−0.23	0.14	.12	.63	> 0.99
SQI	0.08	0.20	−0.32	0.48	.21	.71	> 0.99

Note. Linear regression results for the association between soundscore and sleep parameters among N = 261 Black adults aged 18–85 years living in the Miami metropolitan area. Soundscores are noise estimates derived from HowLoud, ranges from 50 (quiet)–100 (loud). SOL = sleep latency. WASO = nocturnal awakenings. TST = total sleep time. SD = sleep duration, (TST + WASO). SE = sleep efficiency, [TST / (SOL + SD)]. Frequency coupling = fc. REM = rapid eye movement (very-low fc). NREM = non-REM. Light NREM = low fc. Deep NREM = high fc. SQI = sleep quality index (0–100).

^aMeasurements recorded and calculated via SleepImage Ring.

^bLower limit of 95 % CI.

^cUL = upper limit of 95 % CI.

^dHolm-Bonferroni procedure used to correct for multiple hypothesis testing.

^eSimple linear regressions.

^fMultiple linear regression results adjusted for covariates self-reported by participants: age (years), hypertension/depression/anxiety diagnoses, anti-depressant usage, employment status, and income (Likert scale from 1 to 6 in units of $1000: > 10, 10–19, 20–39, 40–59, 60–100, and < 100, respectively).

^*p < .05, two-tailed.

Linear regressions on stratified datasets revealed additional associations between noise and sleep. REM among female participants was found to be negatively associated with soundscore. Though, this relationship did not hold upon adjusting for of covariates (Table 5). Among male participants, noise was observed to be significantly associated with WASO and SD. Greater soundscores were associated with lower reduced WASO and SD. However, only the relationship with WASO was significant following covariate adjustment (Table 6). These same relationships were observed among participants aged 18–54 years, but instead SD not WASO remained significant in adjusted models (Table 7). Conversely, no significant associations were found between soundscore and sleep parameters among participants aged 55–85 years (Table S7). Among the fully stratified analyses, WASO and TST were significantly associated with soundscore for male participants aged 18–54 years and female participants aged 55–85 years, respectively. WASO was shown to be negatively associated with soundscore, and this associations persisted after adjusting for covariates (Table 8). On the other hand, TST was positively associated with soundscore. This relationship was only significant when adjusting for covariates (Table 9). No significant associations between soundscore and sleep parameters were observed among female participants aged 18–54 years and male participants aged 55–85 years (Tables S8 and S9). Lastly, it is important to note that when corrected for multiple comparisons, no associations observed during this investigation remained significant due to the number of sleep parameters assessed.

Table 5.

Linear regression results for the association between soundscore and sleep parameters among female study participants.

Sleep parametera	β	SE	LLb	ULc	$R_{\text{adj}}^2$	p	pcd
Simplee
SOL (s)	−15.67	20.70	−56.51	25.17	< 0.01	.45	.82
WASO (s)	−18.00	22.79	−62.97	26.96	< 0.01	.43	.11
TST (s)	14.58	81.14	−145.54	174.70	< 0.01	.86	.81
SD (s)	−45.44	123.35	−288.86	197.99	< 0.01	.71	.25
SE ( %)	0.26	0.18	−0.10	0.62	.01	.16	.81
Light NREM ( %)	−0.25	0.24	−0.72	0.23	< 0.01	.30	.81
Deep NREM ( %)	0.48	0.30	−0.11	1.08	.01	.11	.81
REM ( %)	−0.23	0.10	−0.43	−0.02	.02	.03*	.82
SQI	0.32	0.24	−0.16	0.80	< 0.01	.19	.81
Multiplef	−3.80	22.61	−48.43	40.83	.05	.87	> 0.99
SOL (s)	−1.79	25.65	−52.42	48.83	.00	.94	.44
WASO (s)	93.60	86.64	−77.44	264.63	.09	.28	.94
TST (s)	42.18	135.14	−224.60	308.96	.05	.76	.44
SD (s)	0.21	0.20	−0.20	0.61	.02	.31	.77
SE ( %)	0.06	0.24	−0.42	0.53	.18	.82	> 0.99
Light NREM ( %)	0.06	0.30	−0.53	0.65	.23	.83	> 0.99
Deep NREM ( %)	−0.11	0.11	−0.33	0.11	.13	.31	> 0.99
REM ( %)	0.05	0.24	−0.43	0.53	.21	.84	> 0.99
SQI	−15.67	20.70	−56.51	25.17	< 0.01	.45	.82

Note. Linear regression results for the association between soundscore and sleep parameters among N = 179 Black females (biological sex assigned at birth) aged 18–85 years living in the Miami metropolitan area. Soundscores are noise estimates derived from HowLoud, ranges from 50 (quiet)–100 (loud). SOL = sleep latency. WASO = nocturnal awakenings. TST = total sleep time. SD = sleep duration, (TST + WASO). SE = sleep efficiency, [TST / (SOL + SD)]. Frequency coupling = fc. REM = rapid eye movement (very-low fc). NREM = non-REM. Light NREM = low fc. Deep NREM = high fc. SQI = sleep quality index (0–100).

^aMeasurements recorded and calculated via SleepImage Ring.

^bLower limit of 95 % CI.

^cUL = upper limit of 95 % CI.

^dHolm-Bonferroni procedure used to correct for multiple hypothesis testing.

^eSimple linear regressions.

^fMultiple linear regression results adjusted for covariates self-reported by participants: age (years), hypertension/depression/anxiety diagnoses, anti-depressant usage, employment status, and income (Likert scale from 1 to 6 in units of $1000: > 10, 10–19, 20–39, 40–59, 60–100, and < 100, respectively).

^*p < .05, two-tailed.

Table 6.

Linear regression results for the association between soundscore and sleep parameters among male study participants.

Sleep parametera	β	SE	LLb	ULc	$R_{\text{adj}}^2$	p	pcd
Simplee
SOL (s)	−12.59	45.73	−103.59	78.41	< 0.01	.78	> 0.99
WASO (s)	−86.68	33.97	−154.28	−19.08	.06	.01*	> 0.99
TST (s)	−182.76	131.38	−444.22	78.70	.01	.17	> 0.99
SD (s)	−421.76	192.18	−804.21	−39.30	.04	.03*	> 0.99
SE ( %)	0.42	0.29	−0.15	0.99	.01	.14	> 0.99
Light NREM ( %)	−0.65	0.41	−1.47	0.17	.02	.12	> 0.99
Deep NREM ( %)	0.81	0.51	−0.20	1.82	.02	.12	.87
REM ( %)	−0.15	0.18	−0.51	0.21	< 0.01	.41	.26
SQI	0.60	0.39	−0.17	1.38	.02	.12	> 0.99
Multiplef
SOL (s)	−30.24	48.77	−127.43	66.96	.03	.54	> 0.99
WASO (s)	−72.87	36.42	−145.45	−0.29	.10	.05*	> 0.99
TST (s)	−202.65	141.63	−484.93	79.63	.04	.16	> 0.99
SD (s)	−417.91	214.00	−844.42	8.59	.01	.05	> 0.99
SE ( %)	0.47	0.29	−0.11	1.05	.14	.11	> 0.99
Light NREM ( %)	−0.12	0.42	−0.95	0.71	.15	.77	> 0.99
Deep NREM ( %)	0.14	0.50	−0.86	1.14	.19	.79	> 0.99
REM ( %)	−0.01	0.18	−0.37	0.35	.15	.97	> 0.99
SQI	0.13	0.39	−0.64	0.90	.18	.74	> 0.99

Note. Linear regression results for the association between soundscore and sleep parameters among N = 82 Black males (biological sex assigned at birth) aged 18–85 years living in the Miami metropolitan area. Soundscores are noise estimates derived from HowLoud, ranges from 50 (quiet)–100 (loud). SOL = sleep latency. WASO = nocturnal awakenings. TST = total sleep time. SD = sleep duration, (TST + WASO). SE = sleep efficiency, [TST / (SOL + SD)]. Frequency coupling = fc. REM = rapid eye movement (very-low fc). NREM = non-REM. Light NREM = low fc. Deep NREM = high fc. SQI = sleep quality index (0–100).

^aMeasurements recorded and calculated via SleepImage Ring.

^bLower limit of 95 % CI.

^cUL = upper limit of 95 % CI.

^dHolm-Bonferroni procedure used to correct for multiple hypothesis testing.

^eSimple linear regressions.

^fMultiple linear regression results adjusted for covariates self-reported by participants: age (years), hypertension/depression/anxiety diagnoses, anti-depressant usage, employment status, and income (Likert scale from 1 to 6 in units of $1000: > 10, 10–19, 20–39, 40–59, 60–100, and < 100, respectively).

^*p < .05, two-tailed.

Table 7.

Linear regression results for the association between soundscore and sleep parameters among study participants aged 18–54 years.

Sleep parametera	β	SE	LLb	ULc	$R_{\text{adj}}^2$	p	pcd
Simplee
SOL (s)	−19.90	20.28	−60.02	20.23	< 0.01	.33	.81
WASO (s)	−56.93	20.81	−98.10	−15.77	.05	.01	.06
TST (s)	−107.07	72.09	−249.69	35.55	.01	.14	.81
SD (s)	−206.15	84.32	−372.96	−39.35	.04	.02*	.12
SE ( %)	0.29	0.12	0.06	0.52	.04	.01*	.12
Light NREM ( %)	−0.34	0.27	−0.86	0.19	< 0.01	.21	.81
Deep NREM ( %)	0.51	0.34	−0.16	1.19	.01	.14	.81
REM ( %)	−0.17	0.12	−0.40	0.06	.01	.15	.81
SQI	0.41	0.27	−0.14	0.95	.01	.14	.81
Multiplef
SOL (s)	−10.03	21.08	−51.76	31.69	.05	.64	> 0.99
WASO (s)	−42.70	22.09	−86.43	1.03	.06	.06	.44
TST (s)	−126.08	75.02	−274.56	22.41	.06	.10	.67
SD (s)	−213.05	88.44	−388.10	−38.00	.07	.02*	.16
SE ( %)	0.19	0.12	−0.05	0.42	.09	.13	.76
Light NREM ( %)	0.05	0.24	−0.43	0.54	.27	.83	> 0.99
Deep NREM ( %)	−0.01	0.31	−0.63	0.60	.29	.97	> 0.99
REM ( %)	−0.03	0.12	−0.27	0.20	.13	.78	> 0.99
SQI	0.06	0.25	−0.44	0.56	.27	.83	> 0.99

Note. Linear regression results for the association between soundscore and sleep parameters among N = 133 Black adults aged 18–54 years living in the Miami metropolitan area. Soundscores are noise estimates derived from HowLoud, ranges from 50 (quiet)–100 (loud). SOL = sleep latency. WASO = nocturnal awakenings. TST = total sleep time. SD = sleep duration, (TST + WASO). SE = sleep efficiency, [TST / (SOL fc). NREM = non-REM. Light NREM = low fc. Deep NREM = high fc. SQI = sleep quality index (0–100).

^aMeasurements recorded and calculated via SleepImage Ring.

^bLower limit of 95 % CI.

^cUL = upper limit of 95 % CI.

^dHolm-Bonferroni procedure used to correct for multiple hypothesis testing.

^eSimple linear regressions.

^fMultiple linear regression results adjusted for covariates self-reported by participants: age (years), hypertension/depression/anxiety diagnoses, anti-depressant usage, employment status, and income (Likert scale from 1 to 6 in units of $1000: > 10, 10–19, 20–39, 40–59, 60–100, and < 100, respectively).

^*p < .05, two-tailed.

Table 8.

Linear regression results for the association between soundscore and sleep parameters among male study participants aged 18–54 years.

Sleep parametera	β	SE	LLb	ULc	$R_{\text{adj}}^2$	p	pcd
Simplee
SOL (s)	−15.16	66.44	−149.68	119.35	< 0.01	.82	> 0.99
WASO (s)	−107.73	42.28	−193.33	−22.13	.12	.02*	.14
TST (s)	−204.11	175.69	−559.78	151.56	.01	.25	.98
SD (s)	−333.25	193.41	−724.79	58.28	.05	.09	.74
SE ( %)	0.43	0.29	−0.15	1.02	.03	.14	.98
Light NREM ( %)	−0.73	0.49	−1.73	0.27	.03	.15	.98
Deep NREM ( %)	0.80	0.67	−0.57	2.16	.01	.24	.98
REM ( %)	−0.06	0.25	−0.56	0.44	< 0.01	.82	> 0.99
SQI	0.71	0.49	−0.29	1.71	.03	.16	.98
Multiplef
SOL (s)	13.72	76.08	−141.45	168.89	< 0.01	.86	> 0.99
WASO (s)	−110.76	45.32	−203.18	−18.34	.20	.02*	.18
TST (s)	−174.72	188.07	−558.29	208.85	.10	.36	> 0.99
SD (s)	−298.94	221.93	−751.58	153.70	.01	.19	> 0.99
SE ( %)	0.39	0.25	−0.13	0.91	.40	.14	> 0.99
Light NREM ( %)	−0.64	0.46	−1.57	0.29	.35	.17	> 0.99
Deep NREM ( %)	0.63	0.59	−0.57	1.84	.40	.29	> 0.99
REM ( %)	0.01	0.24	−0.49	0.51	.21	.97	> 0.99
SQI	0.61	0.43	−0.27	1.49	.41	.17	> 0.99

Note. Linear regression results for the association between soundscore and sleep parameters among N = 40 Black males (biological sex assigned at birth) aged 18–54 years living in the Miami metropolitan area. Soundscores are noise estimates derived from HowLoud, ranges from 50 (quiet)–100 (loud). SOL = sleep latency. WASO = nocturnal awakenings. TST = total sleep time. SD = sleep duration, (TST + WASO). SE = sleep efficiency, [TST / (SOL + SD)]. Frequency coupling = fc. REM = rapid eye movement (very-low fc). NREM = non-REM. Light NREM = low fc. Deep NREM = high fc. SQI = sleep quality index (0–100).

^aMeasurements recorded and calculated via SleepImage Ring.

^bLower limit of 95 % CI.

^cUL = upper limit of 95 % CI.

^dHolm-Bonferroni procedure used to correct for multiple hypothesis testing.

^eSimple linear regressions.

^fMultiple linear regression results adjusted for covariates self-reported by participants: age (years), hypertension/depression/anxiety diagnoses, anti-depressant usage, employment status, and income (Likert scale from 1 to 6 in units of $1000: > 10, 10–19, 20–39, 40–59, 60–100, and < 100, respectively).

^*p < .05, two-tailed.

Table 9.

Linear regression results for the association between soundscore and sleep parameters among female study participants aged 55–85 years.

Sleep parametera	β	SE	LLb	ULc	$R_{\text{adj}}^2$	p	pcd
Simplee
SOL (s)	−2.19	45.53	−92.72	88.35	<0.01	.96	>0.99
WASO (s)	22.46	42.82	−62.70	107.61	<0.01	.60	>0.99
TST (s)	173.92	163.13	−150.48	498.32	<0.01	.29	>0.99
SD (s)	189.79	258.91	−325.07	704.65	<0.01	.47	>0.99
SE ( %)	0.24	0.39	−0.54	1.01	<0.01	.55	>0.99
Light NREM ( %)	−0.24	0.33	−0.90	0.42	<0.01	.48	>0.99
Deep NREM ( %)	0.49	0.42	−0.33	1.32	<0.01	.24	>0.99
REM ( %)	−0.24	0.16	−0.57	0.09	.01	.15	>0.99
SQI	0.26	0.32	−0.38	0.90	<0.01	.43	>0.99
Multiplef
SOL (s)	30.01	54.65	−78.81	138.83	<0.01	.58	>0.99
WASO (s)	42.00	51.63	−60.81	144.80	<0.01	.42	>0.99
TST (s)	379.48	187.60	5.92	753.04	.09	.05*	.42
SD (s)	365.87	317.33	−266.02	997.76	<0.01	.25	>0.99
SE ( %)	0.25	0.49	−0.71	1.22	<0.01	.60	>0.99
Light NREM ( %)	−0.15	0.41	−0.96	0.66	<0.01	.71	>0.99
Deep NREM ( %)	0.23	0.51	−0.78	1.24	<0.01	.65	>0.99
REM ( %)	−0.07	0.19	−0.45	0.32	.06	.73	>0.99
SQI	0.06	0.39	−0.72	0.85	<0.01	.87	>0.99

Note. Linear regression results for the association between soundscore and sleep parameters among N = 86 Black females (biological sex assigned at birth) aged 55–85 years living in the Miami metropolitan area. Soundscores are noise estimates derived from HowLoud, ranges from 50 (quiet)–100 (loud). SOL = sleep latency. WASO = nocturnal awakenings. TST = total sleep time. SD = sleep duration, (TST + WASO). SE = sleep efficiency, [TST / (SOL + SD)]. Frequency coupling = fc. REM = rapid eye movement (very-low fc). NREM = non-REM. Light NREM = low fc. Deep NREM = high fc. SQI = sleep quality index (0–100).

^aMeasurements recorded and calculated via SleepImage Ring.

^bLower limit of 95 % CI.

^cUL = upper limit of 95 % CI.

^dHolm-Bonferroni procedure used to correct for multiple hypothesis testing.

^eSimple linear regressions.

^fMultiple linear regression results adjusted for covariates self-reported by participants: age (years), hypertension/depression/anxiety diagnoses, anti-depressant usage, employment status, and income (Likert scale from 1 to 6 in units of $1000: > 10, 10–19, 20–39, 40–59, 60–100, and < 100, respectively).

^*p < .05, two-tailed.

4. Discussion

4.1. Influence of demographic factors on sleep parameters among Black adults

This study examined the relationship between environmental noise and sleep parameters among Black adults, revealing unique findings with how sleep patterns differ among demographics. Participants aged 18–54 years were observed to have shorter SDs than those aged 55–85 years. This is inconsistent with prior literature, which shows that SD decreases with age [34]. It was also found that participants aged 18–54 years spent less time in REM than participants aged 55–85 years. Again, this opposite of what is typically observed in broader populations [35]. Related, females were shown to have lower REM than males among participants age 18–54 years, which is inverse to normal patterns [36]. In total, these results suggest that age and sex may influence sleep architecture in Black adults in ways distinct from the general population, underscoring the importance of exploring these determinants in this group.

4.2. Associations between environmental noise and sleep parameters

Regression analyses supported the hypothesis that lower environmental noise levels are associated with better sleep health parameters among Black adults. Specifically, quieter environments were associated with less time awake at night and changes in sleep period length. However, findings indicate that the relationship between soundscore and sleep parameters are dependent on the demographic subgroup. For instance, the reduction in nocturnal awakenings due to quieter environments was more pronounced among male participants aged 18–54 years, suggesting a heightened autonomic response to environmental stimuli in this subgroup, potentially driven by hypervigilance. This finding adds nuance to prior studies, which have predominantly shown that women are more susceptible to external sleep disturbances [37]. Several factors could be potentially at the root of this. Hypervigilance is posited to be an adaptive response to experiencing racism and other stressors [38]. Furthermore, it can be an agentic behavior, as Black men report it to be an effective coping strategy for navigating inhibiting environments [39]. Future investigations are warranted to further address this relationship. Another notable finding was that soundscore was observed to have inverse associations between soundscore and length of sleep period. SD was negatively associated with soundscore in participants aged 18–54, while TST was positively associated with soundscore among female participants aged 55–85 years. Although distinct measures, TST and SD are intrinsically related, where SD = TST + WASO. The decrease in SD among male participants could partially be explained by the drop in WASO. The increase in TST among female participants aged 55–85 years observed in quieter environments may indicate that higher noise levels prevent optimal sleep patterns among this demographic. This further provides evidence that women are more likely to be disturbed by noise stimuli [40]. Age appeared to alter the relationship between noise and sleep parameters. Specifically, lower noise levels were linked to longer sleep times in older participants but lower sleep time in younger ones. This could possibly be due to age-related changes in the auditory system, which can alter sensitivity to sound stimuli [41]. This finding suggests that while younger adults may respond strongly to environmental noise with disrupted sleep, older adults might experience autonomic arousal at lower sound thresholds, thus rendering their sleep vulnerable even in relatively quieter environments. These age-specific findings emphasize the need to consider sensory changes over the lifespan when assessing environmental influences on sleep.

4.3. Direct associations between noise and sleep

The relationships between environmental noise and sleep parameters found in this study indicate that among Black adults, there exists variability in physiological responses induced by exposure to noise. This could stem from differences in noise sensitivity impacting the levels of autonomic arousal from sound disturbances. Noise sensitivity may amplify these direct effects by heightening physiological responses and even moderating noise levels. For example, female participants aged 55–85 years, exhibited more significant changes in time slept in response to noise levels, possibly due to age-related changes in auditory sensitivity. This heightened sensitivity implies that noise exposure directly impacts the quality of sleep by triggering arousal responses, limiting the ability to enter restorative stages of sleep, and subsequently leading to more fragmented sleep.

4.4. Indirect associations between noise and sleep

Beyond direct physiological responses, the observed relationship between noise exposure and poorer sleep health among Black adults may also reflect broader socioeconomic and environmental trends. Black individuals are disproportionately likely to live in high-noise areas, often due to systemic factors that limit residential options to urban environments with high levels of noise pollution. Many areas with high populations of Black individuals are near major noise sources such as highways, industrial zones, and densely populated urban centers [42]. These environments not only increase noise exposure but also come with additional stressors, including higher levels of air pollution, traffic congestion, and reduced access to green spaces, all of which can indirectly impact sleep quality [43]. Living near highways or in areas with constant urban noise may perpetuate a cycle of sleep deprivation, as constant exposure to noise-related disturbances contributes to baseline stress levels and impacts mental and physical health. Additionally, individuals in such neighborhoods often face additional socioeconomic disadvantages, such as limited access to healthcare, fewer resources for stress management, and higher exposure to daily psychosocial stressors [44]. These combined factors can increase the vulnerability to sleep disturbances beyond noise alone as financial stress, crowded living conditions, and lack of restorative outdoor spaces add layers of strain. In this context, noise exposure may be a marker for the cumulative burden of socioeconomic and environmental stressors that impair sleep health. Addressing these sleep disparities in Black communities, therefore, may require a holistic approach that not only aims to reduce noise pollution but also tackles the broader inequities driving these conditions, such as residential segregation and economic inequality.

4.5. Limitations

While this study provides novel insights, certain limitations should be noted. The use of HowLoud as a measure of environmental noise, although objective and reliable, may not fully capture individual noise exposure. For instance, participants in well-insulated homes may experience less noise intrusion than estimated by outdoor noise levels. Additionally, sleep stages were measured using a consumer-grade wearable device rather than PSG, the gold standard for sleep assessment [45]. This may limit the precision of findings related to sleep architecture. Furthermore, effect sizes for the regression analyses were small, and the statistical power was reduced in analyses involving male participants due to their lower sample size. Related to this, results were insignificant when corrected for multiple comparisons. However, this is likely due to the number of sleep parameters assessed during this investigation. Future studies should seek to validate these findings using larger samples and polysomnographic measurements to explore further the role of psychosocial and biological factors in the relationship between noise and sleep physiology among Black adults.

5. Conclusion

To the authors’ current knowledge, the current study was among the first to investigate the associations between neighborhood noise levels and sleep architecture in Black adults using strictly objective measures. This study highlights the complex interplay between environmental noise and sleep architecture among Black adults, with findings suggesting that the impact of noise on sleep varies across demographic subgroups. The observed associations between noise and specific sleep parameters, such as deep and REM sleep, suggest that noise sensitivity and autonomic response profiles are influenced by factors like age, sex, and individual sleep needs. These results underscore the importance of considering both environmental and demographic factors when assessing sleep health disparities. They provide a foundation for future research to understand and mitigate the impact of environmental noise on sleep in Black communities.

Supplementary Material

Supplementary materials

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.sleepe.2025.100116.

Availability of data

Participant data used in this study are not publicly available.