Association Between Community Noise and Adiposity in Patients With Cardiovascular Disease

Abstract

Introduction:

This study aimed to explore the effect of community noise on body mass index (BMI) and waist circumference (WC) in patients with cardiovascular disease (CVD).

Materials and Methods:

A representative sample of 132 patients from three tertiary hospitals in the city of Plovdiv, Bulgaria was collected. Anthropometric measurements were linked to global noise annoyance (GNA) based on different residential noise annoyances, day–evening–night (L _den), and nighttime (L _night) road traffic noise exposure. Noise map L _den and L _night were determined at the living room and bedroom façades, respectively, and further corrected to indoor exposure based on the window-opening frequency and soundproofing insulation.

Results and Discussion:

Results showed that BMI and WC increased (non-significantly) per 5 dB. The effect of indoor noise was stronger in comparison with that of outdoor noise. For indoor L _den, the effect was more pronounced in men, those with diabetes, family history of diabetes, high noise sensitivity, using solid fuel/gas for domestic heating/cooking, and living on the first floor. As regards indoor L _night, its effect was more pronounced in those with low socioeconomic status, hearing loss, and using solid fuel/gas for domestic heating/cooking. GNA was associated with lower BMI and WC.

Conclusion:

Road traffic noise was associated with an increase in adiposity in some potentially vulnerable patients with CVD.

Introduction

Recently, evidence of the metabolic effects of environmental noise has been emerging. Several studies explored the impact of road traffic noise on adiposity among adults,[1,2,3,4,5,6] youth,[7] and children,[8] but the associations were generally small and inconsistent. The underlying biomedical mechanism refers to noise acting as a stressor, involving the stimulation of the hypothalamic-pituitary-adrenal and simpatico-adrenal axes and resulting in increase in catecholamine, cortisol, and oxidative stress levels, which could engender, in the long run, insulin resistance and accumulation of adipose tissue.[9,10] At the same time, nighttime noise causes sleep disturbances[9] which are associated with adiposity through modulation of the cortical appetite control mechanisms (reduction of leptin and elevation of ghrelin levels).[11,12,13] It was even suggested that maternal exposure during pregnancy might raise the risk of overweight of their offspring.[8]

Another, behavioral pathway was proposed by Foraster et al.[14] They observed lower level of physical activity with increasing noise annoyance.[14] Roswall et al. recently extended these findings to modelled road traffic noise, which was also found to be negatively associated with leisure-time sports.[15] This effect is possibly due to impaired sleep quality, leading to higher daytime sleepiness and decreased willingness to engage in physical activity.[14] Another angle to this would be to view noise not merely as stressor, but also as a constraint on restorative qualities of the living environment, thereby leading to decreased residential satisfaction,[16] which could entail less engagement in outdoor physical activity.[17,18]

No study has so far looked into the relationship noise − adiposity in vulnerable subgroups such as patients with pre-existing cardiovascular disease (CVD). This could be of relevance to cardiologists for two reasons. On one hand, adiposity is not only a risk factor for incident CVD,[19] but for developing coronary artery disease among patients with pre-existing essential hypertension,[20] and for mortality among patients with coronary artery disease.[21] On the other, it is highly prevalent and inadequately managed in patients. EUROASPIRE IV, a cross-national European study, showed that 59.9% of the coronary artery disease patients in 2013 were physically inactive, 37.6% were obese, and 58.2% were centrally obese.[22] In fact, these proportions had increased with some 7% since 1999.[23] Clearly, targeting classic risk factors for adiposity like physical inactivity and poor dietary habits is not sufficient. Therefore this study aimed to explore the association of community noise exposure with body mass index (BMI) and waist circumference (WC) among Bulgarian patients with pre-existing CVD.

Materials and Methods

Data collection

This was a cross-sectional study conducted in the city of Plovdiv, Bulgaria (March–May 2016). Detailed description of the design is reported elsewhere.[24] Briefly, patients admitted in three tertiary hospitals were enrolled in the study. A cardiologist/internist took their medical history, reviewed their medical documentation and medication, and performed anthropometric measurements. Patients filled out a questionnaire asking about sociodemographic, lifestyle, socio-acoustic, housing, and medical factors. The questions were initially piloted in a sample of 20 patients. The study was reviewed and approved by the Ethics Review Committee at Medical University of Plovdiv. It adhered to the Declaration of Helsinki, participants signed informed consent declarations, and their participation was voluntary and anonymous. Participants received no incentives.

Inclusion/exclusion criteria

We included Bulgarian-speaking, capacitated, and literate patients, aged over 18 years, and resident in the Plovdiv Province during the preceding 12 months. Requirement was that participants were diagnosed with hypertension (ICD-10: I10–I13) and/or ischemic heart disease (ICD-10: I20–I25) at least 12 months prior to their current hospitalization. Exclusion criteria were as follows: secondary hypertension (ICD-10: I15), secondary diabetes (ICD-10: E08, E09, E13), cancer, and pregnancy.

Outcome measures

BMI (weight in kg divided by height squared in cm) and WC were used as proxies for adiposity. Body weight was measured in light clothing on a calibrated digital scale. Height was measured standing without shoes. WC was measured at the midpoint between the lower margin of the last palpable rib and the top of the iliac crest, using a stretch-resistant tape. Measurements were performed twice at the end of a normal expiration.[25]

Noise exposure

We used the strategic noise map of Plovdiv (according to the Environmental Noise Directive 2002/49/EC) to assign modelled noise exposure to each participant living in the city (n = 132). It shows 5-dB noise contours for road traffic day–evening–night equivalent noise level (L _den) and night equivalent noise level (L _night).[26] Participants’ addresses were geocoded manually − they were inspected one-by-one using 3D Google Earth imagery, Google Street View, and in-person visits. The questionnaire asked about the orientation of rooms within the dwelling, which made it possible to assign L _den and L _night to the living room and bedroom façades, respectively.[24] This façade noise exposure was corrected to indoor exposure using the method proposed by Foraster et al.[27] An insulation factor was subtracted from the façade noise level depending on the window-opening frequency (−15 dB for “always,” −20 dB for “sometimes,” and −30 dB for “never” keeping the living room windows open) and the presence of a soundproofing insulation (−40 dB for “never” and having an insulation). L _den and L _night were treated as continuous (per 5 dB) variables.

Noise annoyance

Participants reported their annoyance by different noise sources: “To what extent, on a scale from 0 to 10, are you disturbed, annoyed or irritated by noise from different sources while you are at home (IN THE PAST 12 MONTHS)?” (“0, not at all” to “10, very much”).” Annoyances due to traffic, industrial, neighborhood (playgrounds, recreational facilities, etc.), building (e.g., plumbing, elevators, etc.), and dwelling/apartment noise (appliances, toddlers, etc.) were averaged on a global noise annoyance (GNA) scale (0–10).

Covariates

The questionnaire asked about gender, age, ethnicity, highest educational attainment, marital status, self-rated socioeconomic status (SES), diet quality, alcohol drinking, cigarette smoking, physical activity, hearing loss, noise sensitivity, energy used for domestic heating/cooking, floor of the dwelling/apartment, duration of residence, and the average time/day spent at home. Participants’ diet quality was assessed with a food frequency questionnaire. Their family history of ischemic heart disease, stroke, and diabetes, and their co-morbidity with ischemic heart disease (ICD-10: I20–I25), stroke (ICD-10: I61, I63–I64), hypertension (ICD-10: I10–I13), and diabetes (ICD-10: E10–E12) were determined as well. Exposure to traffic fine particulate matter (PM_2.5) was modeled. For more details on these covariates, refer Dzhambov et al.[24]

Statistical analyses

Missing values were replaced using the expectation-maximization algorithm. The distribution of the dependent variables was explored with the D’Agostino–Pearson K ² test and histograms. Outliers were winsorized (recoded closer to the nearest non-outlier value).

The bivariate associations between road traffic noise exposure, GNA, and other variables were examined with Welch’s t-test/analysis of variance, Pearson’s chi-squared test/Fisher’s exact test/Fisher–Freeman–Halton test, and the Spearman correlation.

The multivariate association between road traffic noise and GNA and patients’ BMI and WC was explored using mixed linear models with a restricted maximum likelihood estimator and a random intercept to account for clustering on hospitals. Four adjusted models were fitted based on a priori selection of important covariates using DAGitty v. 2.3 (http://www.dagitty.net/).

Potential effect modification by participants’ characteristics was tested using the Wald test for interactions at the relaxed P < 0.2 level,[28,29] and the method of Altman and Bland.[30] Other results were considered statistically significant at the P < 0.05 (two-tailed) level.

Results

Detailed description of the sample, associations between different variables in the dataset, and correlations between the exposure indicators are reported in Dzhambov et al.[24] Briefly, mean age was 62.30 ± 14.61 years and 48.5% (n = 64) were men. L _den was significantly associated with male gender, sleep disturbance, floor of the dwelling/apartment, duration of residence, time spent at home/day, and traffic PM_2.5. The mean BMI and WC in the sample were 29.55 ± 5.76 kg/m² and 94.71 ± 19.96 cm, respectively. Higher BMI was significantly associated with lower SES (r _s = −0.14, P = 0.035), quiet bedroom orientation (r _s = −0.15, P = 0.027), solid fuel/gas used for domestic heating/cooking (r _s = 0.20, P = 0.003), and lower floor of the dwelling/apartment (r _s = −0.19, P = 0.006). Higher WC was associated with male gender (r _s = −0.23, P = 0.001), solid fuel/gas used for domestic heating/cooking (r _s = 0.15, P = 0.033), and lower floor of the dwelling/apartment (r _s = −0.20, P = 0.003). In comparison with nationally representative data for CVD patients, most clinical variables in this study were representative.[31,32]

Table 1 shows the adjusted linear models for road traffic noise. There was a non-significant increase in BMI and WC per 5-dB. The effect of indoor noise exposure was more pronounced. GNA was associated with a significantly lower BMI and WC [Table 2].

Table 1.

Associations between road traffic noise (per 5 dB) and adiposity (mixed linear models) (n = 132)

Exposure indicator	Change (95% CI)
	Body mass index (kg/m2)	Waist circumference (cm)
Outdoor L den
Basic model (gender and age)	−0.04 (−0.85, 0.77)	0.68 (−2.16, 3.52)
Main modela	−0.15 (−1.01, 0.72)	0.10 (−2.72, 2.92)
Indoor L den
Basic model (gender and age)	0.14 (−0.53, 0.81)	1.89 (−0.44, 4.22)
Main modela	0.03 (−0.72, 0.78)	1.34 (−1.09, 3.78)
Outdoor L night
Basic model (gender and age)	−0.16 (−0.97, 0.64)	−0.01 (−2.82, 2.79)
Main modela	0.05 (−0.83, 0.92)	0.95 (−1.90, 3.80)
Indoor L night
Basic model (gender and age)	0.13 (−0.41, 0.67)	0.77 (−1.12, 2.65)
Main modela	0.21 (−0.39, 0.81)	1.14 (−0.81, 3.09)

L _den = day–evening–night equivalent noise level, L _night = night equivalent noise level. ^aAdjusted for gender, age + ethnicity (“Bulgarian” or “other”), socioeconomic status (“low” or “middle/upper”), education (“primary/junior high school or less,” “secondary,” or “higher”), diet quality (continuous), smoking (“never,” “former,” or “current smoker”), alcohol drinking (“lifetime abstainer/former drinker,” “current light drinker,” or “current moderate/heavy drinker”), physical activity (“inactive,” “low activity,” or “active”), noise sensitivity (continuous), and fine particulate matter (“0.0–2.0 μg/m³” or “2.0–34.94 μg/m³”).

Table 2.

Associations between global noise annoyance (per one interquartile range) and adiposity (mixed linear models) (n = 132)

Exposure indicator	Change (95% CI)
	Body mass index (kg/m2)	Waist circumference (cm)
Global noise annoyance
Basic model (gender and age)	−1.91 (−3.22, −0.59)*	−4.71 (−9.38, −0.05)*
Main modela	−2.34 (−4.27, −0.41)*	−7.89 (−14.19, −1.59)*

L _den = day–evening–night equivalent noise level, L _night = night equivalent noise level. Interquartile range for global noise annoyance is 2.80. Change is significant at *P < 0.05. ^aAdjusted for gender, age + ethnicity (“Bulgarian” or “other”), socioeconomic status (“lower,” or “middle/upper”), education (“primary/junior high school or less,” “secondary,” or “higher”), diet quality (continuous), smoking (“never,” “former,” or “current smoker”), alcohol drinking (“lifetime abstainer/former drinker,” “current light drinker,” or “current moderate/heavy drinker”), physical activity (“inactive,” “low activity,” or “active”), noise sensitivity (continuous), and fine particulate matter (“0.0–2.0 μg/m³” or “2.0–34.94 μg/m³”).

In sensitivity analysis, we checked for effect modification by participants’ characteristics. [Table 3] The effect of indoor L _den was more pronounced in men, those with diabetes, family history of diabetes, high noise sensitivity, using solid fuel/gas for domestic heating/cooking, and living on the first floor. A statistically significant effect was found in those over 63 years of age, highly sensitive to noise, with a living room facing a noisy street, and whose apartment was on the first floor. The effect of indoor L _night was more pronounced in those with low SES, hearing loss, and using solid fuel/gas for domestic heating/cooking, and it was statistically significant in those over 63 years of age, reporting low SES, hearing loss, and living on the first floor.

Table 3.

Stratified analysis for identifying potential effect modification of the association between indoor road traffic noise (per 5 dB) and waist circumference (mixed linear models)

	Change (95% CI) (cm)
	Indoor L den	p interaction	Indoor L night	p interaction
Gender		0.197		0.452
Men	2.97 (−0.50, 6.43)		1.93 (−1.01, 4.86)
Women	−0.55 (−4.63, 3.53)		0.42 (−2.20, 3.04)
Age		0.547		0.229
<63 years	1.50 (−2.03, 5.03)		0.75 (−2.11, 3.61)
≥63 years	2.90 (0.04, 5.76)*		3.05 (0.63, 5.47)*
Socioeconomic status		0.276		0.070
Low	5.46 (−2.25, 13.16)		5.09 (0.41, 9.78)*
Middle/upper	0.89 (−1.99, 3.77)		0.35 (−1.75, 2.45)
Cardiovascular disease		0.284		0.701
No	0.20 (−3.62, 4.01)		1.43 (−1.30, 4.17)
Yes	3.15 (−0.67, 6.97)		2.28 (−1.08, 5.63)
Diabetes		0.127		0.800
No	−0.92 (−4.46, 2.62)		0.47 (−2.53, 3.47)
Yes	3.08 (−0.63, 6.79)		0.97 (−1.48, 3.42)
Body mass index		0.661		0.421
<30 kg/m2	1.89 (−0.16, 3.95)		0.66 (−0.89, 2.20)
≥30 kg/m2	2.82 (−0.79, 6.43)		2.22 (−1.24, 5.69)
Family history of cardiovascular disease		0.834		0.376
No	0.60 (−3.03, 4.23)		2.26 (−0.41, 4.93)
Yes	1.16 (−2.60, 4.93)		0.53 (−2.22, 3.28)
Family history of diabetes		0.051		0.530
No	−2.78 (−6.40, 0.85)		1.58 (−1.53, 4.68)
Yes	2.65 (−1.43, 6.73)		0.26 (−2.45, 2.97)
Hearing loss		0.367		0.115
No	0.66 (−3.40, 4.71)		0.52 (−2.40, 3.44)
Yes	2.98 (−0.004, 5.97)		3.67 (1.06, 6.29)*
Noise sensitivity		0.072		0.426
<5	0.69 (−2.40, 3.78)		1.46 (−0.83, 3.76)
≥5	5.07 (1.43, 8.71)*		−0.36 (−4.20, 3.49)
Bedroom orientation		0.600		0.600
Quiet façade	0.68 (−2.71, 4.07)		1.85 (−0.77, 4.47)
Noisy façade	2.20 (−2.36, 6.76)		3.06 (−0.61, 6.73)
Living room orientation		0.288		0.962
Quiet façade	2.60 (−1.17, 6.36)		2.05 (−0.69, 4.79)
Noisy façade	5.72 (0.43, 11.01)*		2.15 (−0.79, 5.09)
Heating/cooking energy		0.232		0.186
Electricity/steam radiator only	0.21 (−2.65, 3.07)		0.58 (−1.49, 2.65)
Wood/coal/gas used	3.77 (−1.32, 8.86)		3.61 (−0.38, 7.59)
Floor of the dwelling		0.052		0.502
1st	4.43 (1.86, 6.99)*		2.30 (0.22, 4.38)*
>1st	−0.14 (−3.97, 3.69)		1.02 (−2.08, 4.12)
Duration of residence		0.657		0.257
≤30 years	0.72 (−2.92, 4.37)		−0.47 (−3.00, 2.07)
>30 years	1.89 (−1.76, 5.55)		2.06 (−1.51, 5.62)
Time at home/day		0.256		0.549
<12 h	−0.41 (−4.86, 4.03)		0.65 (−2.76, 4.06)
≥12 h	2.84 (−0.57, 6.25)		1.98 (−0.72, 4.68)
PM2.5		0.401		0.961
<2.0 μg/m3	−0.43 (−5.46, 4.60)		1.54 (−2.69, 5.78)
>2.0 μg/m3	1.78 (−0.84, 1.40)		1.66 (−0.43, 3.75)

L _den = day–evening–night equivalent noise level, L _night = night equivalent noise level, PM_2.5 = fine particulate matter. Models were adjusted for gender, age, ethnicity, socioeconomic status, education, diet quality, smoking, alcohol drinking, physical activity, and noise sensitivity. (When the model was stratified by the respective factor, it was removed from the covariate set.) Change is significant at *P < 0.05.

Discussion

Key findings

This study examined the association between road traffic noise and GNA and adiposity in patients with CVD. Road traffic noise was positively associated with BMI and WC in the total sample, but the estimates failed statistical significance. Indoor noise exposure was associated with a greater increase in BMI and WC.

Sensitivity analysis suggested that men, those with diabetes, with family history of diabetes, highly sensitive to noise, using solid fuel/gas for domestic heating/cooking, and living on the first floor were more vulnerable to indoor L _den. The effect of indoor L _night was more pronounced in those with low SES, hearing loss, and using solid fuel/gas for domestic heating/cooking. GNA was associated with lower BMI and WC.

Although previous studies on traffic noise and obesity were population-based, our findings are in line with theirs. Pyko et al. reported 0.21 cm [95% confidence interval (CI): 0.01, 0.41] change in WC and −0.08 kg/m² (95% CI: −0.16, 0.01) in BMI per 5 dB.[4] Oftedal et al. found no effect on BMI or WC per 10 dB.[1] In the study of Christensen et al., there was 0.25 cm (95% CI: 0.11, 0.39) increase in WC and 0.17 kg/m² (95% CI: 0.12, 0.22) in BMI per 10 dB increase of 1-year-preceding noise levels.[3] In a longitudinal study, Christensen et al. estimated a gain of 0.17 cm (95% CI: −0.036, 0.38) in WC per 10 dB for the 5 years preceding follow-up.[2] As for air traffic noise, authors found 1.51 cm (95% CI: 1.13, 1.89) increase in WC and 0.04 kg/m² (95% CI: −0.05, 0.13) in BMI for every 5 dB.[5] This suggests that there may not be a significant difference in noise effects between people from the general population and patients with CVD. Our models for indoor noise exposure yield much larger effect size estimates, and this could not be compared to previous studies.

We found stronger association for indoor L _den in men, whereas most previous studies did not report significant gender differences.[2,3,4] Still, Oftedal et al. reported greater odds of obesity in men albeit non-significant.[1] In Eriksson et al., there was a significantly higher increase in WC per 5 dB in men (2.26 cm, 95% CI: 1.83, 2.69) versus women (1.58 cm, 95% CI: 1.13, 2.03).[5]

With respect to age, only Pyko et al. evidenced a significant effect modification, with those under 60 years being at higher risk.[4] In our study, the effect was stronger and statistically significant in people over 63 years although the interaction itself was not significant.

Herein, indoor L _den had a stronger effect in highly noise sensitive participants, as in Oftedal et al.’s work.[1] Family history of diabetes was an effect modifier for indoor L _den in our study but not in previous studies.[4,5]

In contrast to Christensen et al.,[3] in our study there was a more pronounced association for indoor L _night in people with low SES.

For people living on the first floor, the effect of indoor L _den was significantly stronger, possibly because the sound propagation model used for noise calculations assumed a receiver point at 4 m above the terrain level. Thus, for those living on higher floors the potential for exposure misclassification is expected to be greater. No other study has tested for effect modification by indoor air pollution or hearing impartment, both of which seemed to aggravate the effect of noise in ours.

Biomass fuel-combustion smoke deserves consideration as a co-exposure. Tentative evidence suggests a link between outdoor air pollution and obesity in children[33] and adults.[34] However, to our knowledge, no study has so far linked indoor air pollution to markers of adiposity or tested its interaction with road traffic noise. Given the potential of specific air contaminants to induce systemic inflammation in the body and modulate leptin levels[35] and insulin resistance,[36] such an association seems plausible and could explain the observed effect modification in our study.

Previous studies have not investigated the association between noise annoyance and adiposity. We speculate that the lack of an effect in this study is due to several reasons. First, GNA may not be a good exposure indicator because people may not be able to accurately average their annoyance over a long period of time, needed to observe an effect on adiposity. Also, GNA refers to noise during the waking hours, whereas both L _den and L _night take into account nighttime exposure during sleep. An inverse association was previously reported between annoyance and blood pressure, with significantly higher systolic blood pressure and higher prevalence of hypertension in less annoyed participants; this counterintuitive finding was explained “with the much higher adaptive efforts that higher annoyed subjects invested to reduce noise exposure compared with less annoyed subjects.”[37] Moreover, in the study of Oftedal et al., there was some increase in the effect on WC after it was adjusted for noise annoyance, also suggesting an inverse relationship.[1]

Strengths and limitations

This was the first study to examine the relationship between community noise and adiposity in CVD patients who are typically either excluded from the analyses or simply analyzed as a subgroup. Thus, out findings suggest that noise may be important not only to Bulgarian public health officers, but to clinicians and general practitioners as well, because 43% of the patients with ischemic heart disease in Bulgaria are obese, 57% have central obesity, and 74% are physically inactive.[38] Another strength of the study is the fact that we had detailed exposure data, including the location of rooms and indoor noise exposure, which showed good validity with respect to hypertension in the studies of Babisch et al.[39] and Foraster et al.,[27] respectively. Our results also indicated stronger effect of indoor noise (in comparison with façade-level noise) which might explain the inconclusive results of previous studies in the field. Finally, we gathered data on multiple confounders and effect modifiers.

There are several important limitations. First, the study was cross-sectional precluding causal inference. This was addressed by running a time window analysis. Oftedal et al.[1] found stronger (but not significantly so) effect among long-term residents. Eriksson et al. reported significantly stronger effect on WC among non-movers during the study period.[5]

The sample size was smallish although justified by an a priori simulation.[24] This fact may account for some non-significant effects. We had to relax the P-value for interaction terms to have more power. Moreover, we lacked power to dichotomize BMI and WC, which could be important if noise affects the right side of the distribution of the measures of adiposity more than its mean.[8]

The limitations of the noise maps and air pollution models for Plovdiv are discussed elsewhere.[40] We refined the noise assessment method but the END noise maps have well-known issues − for example, they often use traffic count data only for major roads, free sound propagation, and do not consider the shielding of buildings, which leads to exposure misclassification.[41]

We did not test whether sleep disturbance and physical activity mediated the effect of road traffic noise on adiposity. However, this is an issue that may be addressed using structural equation modeling in future studies in order to disentangle the mechanisms underlying the observed associations.

Some studies show a correlation between the distribution of urban traffic noise and social inequalities.[42,43] Therefore, critics may argue that the effect on adiposity observed in our study is due to the fact that patients living in more disadvantaged areas were also exposed to higher noise levels and that, in fact, social deprivation was the leading adverse factor. Still, the picture is not so clear, and this spatial correlation appears to be region-specific. For instance, one study indicated higher noise exposure in advantaged neighbourhoods[44] and another one, a non-linear relationship between noise and SES.[45] Another related limitation is that we lacked aggregated spatial data to construct a neighborhood-level SES index. There are no Bulgarian studies on this subject, but reassuringly, there was no correlation between individual-level SES and road traffic noise in our study. Moreover, in Bulgaria, the scatter of people of diverse SES in the same residential area or even at the same street is common due to both cultural and social factors. It is more than likely that neighborhood selection among elderly CVD patients, who face many daily challenges due to the high poverty rates and inadequate access to healthcare in Bulgaria, is driven by other factors in contrast to more affluent countries.

Initially, when we piloted the preliminary, lengthier version of the questionnaire, we observed lower compliance and quality of patients’ responses (e.g., omitted or implausible responses). This prompted us to reduce the total number of questions and to use fairly simplistic measures for some covariates, for instance, noise sensitivity. This could have biased the results since single-item ratings of noise sensitivity have notable shortcomings compared to multi-item questionnaires, such as crude precision, lower retest reliability, and undesired correlations with noise exposure levels.[46] In our sample, noise sensitivity was related to L _night (r _s = 0.26–0.28).

We cannot rule out the possibility that using solid fuel/gas for domestic heating/cooking instead of electricity, which is more expensive, acted as a proxy for low SES and household deprivation. This is also implied by the correlation between this variable and SES (r _s = −0.24). We attempted to address this issue by adjusting for SES, education, and ethnicity.

Finally, other noise sources, including occupational noise, were not considered, since very few people were living close to a railway or reported exposure to noise at work to allow for meaningful analysis.

Conclusion

This study showed evidence that road traffic noise was associated with an increase in adiposity in some potentially vulnerable patients with CVD. Indoor noise exposure was a better predictor of BMI and WC. Men, diabetics, those with family history of diabetes, high noise sensitivity, using solid fuel/gas for domestic heating/cooking, and living on the first floor were more vulnerable to indoor L _den. The association for indoor L _night was stronger for those with low SES, hearing loss, and using solid fuel/gas for domestic heating/cooking. GNA was associated with lower adiposity.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.