Chang Ve Ark. 2013 Occupational Noise Exposure and Incident Hypertension in Men A Prospective
Chang Ve Ark. 2013 Occupational Noise Exposure and Incident Hypertension in Men A Prospective
Chang Ve Ark. 2013 Occupational Noise Exposure and Incident Hypertension in Men A Prospective
8
© The Author 2013. Published by Oxford University Press on behalf of the Johns Hopkins Bloomberg School of DOI: 10.1093/aje/kws300
Public Health. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. Advance Access publication:
March 6, 2013
Original Contribution
Ta-Yuan Chang*, Bing-Fang Hwang, Chiu-Shong Liu, Ren-Yin Chen, Ven-Shing Wang,
Initially submitted March 19, 2012; accepted for publication June 25, 2012.
The associations between occupational noise exposure and hypertension remain controversial because of the
differences in study designs, exposure assessments, and confounding controls. This prospective study investi-
gated the relationship between noise exposure and the 10-year risk of hypertension. A cohort of 578 male
workers in Taiwan was followed from 1998 to 2008. All subjects were divided into high-, intermediate-, and low-
exposure groups on the basis of noise exposure assessment. Cox regression models were used to estimate the
relative risks of hypertension after adjustment for potential confounders. During the 7,805 person-years of follow-
up, 141 hypertension cases were identified. Significant increases of 3.2 (95% confidence interval (CI): 0.2, 6.2)
mm Hg in systolic blood pressure and 2.5 (95% CI: 0.1, 4.8) mm Hg in diastolic blood pressure between the
baseline and follow-up measurements were observed in the high-exposure group. Participants exposed to ≥85
A-weighted decibels (dBA) had a 1.93-fold (95% CI: 1.15, 3.22) risk of hypertension compared with those
exposed to <80 dBA. There was a significant exposure-response pattern (P = 0.016) between the risk of hyper-
tension and the stratum of noise exposure. Prolonged exposure to noise levels ≥85 dBA may increase males’
systolic and diastolic blood pressure levels. This association may translate into a higher incidence of
hypertension.
Abbreviations: CI, confidence interval; dBA, A-weighted decibel(s); DBP, diastolic blood pressure; RR, relative risk; SBP, sys-
tolic blood pressure; SD, standard deviation.
Chronic exposure to noise has been associated with car- risk of hypertension in 2 retrospective cohort studies (10, 11),
diovascular disease, including ischemic heart disease (1), 2 repeated-measure studies (12, 13), and 9 cross-sectional
myocardial infarction (2–5), coronary heart disease (6, 7), studies (14–22). However, the results of other studies are
and stroke (8). This association may exist because noise inconsistent with these findings (23–31). Reasons for this
exposure activates the sympathetic and endocrine systems inconsistency may include differences in study design, dif-
to affect the humoral and metabolic states of the human ferences in exposure assessment, different degrees of
organism, producing the increase in blood pressure and the ability to control for potential confounders, and various
changes in other biological risk factors (such as blood degrees of the use of hearing-protective devices at work.
lipids and glucose levels) that promote the development of One cohort study reported a relationship between occu-
hypertension and cardiovascular diseases (9). pational noise exposure and the incidence of hypertension
The existence of an association between noise exposure (10). However, these results were limited by an exposure
in occupational settings and hypertension is still controver- bias caused by no adjustments for the use of hearing-
sial. Occupational noise exposure has been associated with protective devices, and the association between noise expo-
a sustained elevation of blood pressure or with a higher sure and blood pressure was not reported. In addition,
important risk factors for hypertension, such as body mass factors included age, educational level, employment dura-
index, cigarette use, alcohol intake, regular exercise, salt tion, cigarette use, alcohol intake, regular exercise, the use
intake, and a family history of hypertension (32, 33), were of antihypertensive medication, and the use of hearing-
not considered. The objective of this study was to investi- protective devices. Additional information (such as salt
gate the relationship between prolonged exposure to occu- intake and a family history of hypertension) was included
pational noise and the 10-year incidence of hypertension only in 2008. To avoid information bias, we defined the
by taking these important factors into account. lifestyle habits for regular users specifically (21, 34). The
use of hearing-protective devices included the percentage of
MATERIALS AND METHODS
time that the subjects wore hearing-protective devices (i.e.,
never use, <2 hours’, 2–4 hours’, >4–6 hours’, and >6–8
Study population hours’ working time) and the type of hearing-protective
devices (i.e., earplugs, earmuffs, or both). High salt-intake
This study was conducted by performing a follow-up workers were defined as those who reported ingesting food
study to a cross-sectional survey in an aircraft manufacturing
Blood pressure measurements and definition of The procedure for noise exposure assessment in 1998
hypertension was similar to that for the follow-up measurements in
2008. We first identified 18 departments in this company
The procedure for baseline blood pressure measurements and divided each department into different locations on the
in 1998 was the same as that for the follow-up measurements basis of the manufacturing processes by industrial hygien-
in 2008. All subjects were required to fast overnight before ists and senior workers. After the walk-through survey, we
blood sampling and blood pressure measurements during measured the 15-minute time-weighted average equivalent
annual health examinations. Subjects sat for 10 minutes in a sound level by using a sound analyzer (Model TES-1358;
chair with back support before blood pressure was measured TES Electronic Corp., Taipei, Taiwan), which was calibrat-
bilaterally by a trained nurse using an automated sphygmo- ed with a sound-level calibrator (Model TES-1356; TES
manometer (Ostar Model P2; Ostar Meditech Corp., Taipei, Electronic Corp.) before environmental monitoring. The
Taiwan). The mean value of the 2 measurements was recorded short-term environmental sampling was performed at 337
to represent the individual’s blood pressure in the present locations that were possibly the loudest workplaces at this
study. Subjects were defined as hypertensive if they reported company. For the 121 sites exhibiting a 15-minute time-
a diagnosis of hypertension given by physicians after 1998, weighted average equivalent sound level of ≥65 A-weighted
if the mean value of their resting systolic blood pressure decibels (dBA), additional 8-hour time-weighted average
(SBP) was ≥140 mm Hg in 2008, or if the mean value of measurements were conducted during September–December
their resting diastolic blood pressure (DBP) was ≥90 mm Hg of 1998. All subjects were divided into one of similar expo-
in 2008. Height, body weight, total cholesterol level, and tri- sure groups on the basis of the similarity and frequency
glyceride level were also measured in all subjects at baseline of tasks performed, the agents and processes with which
and follow-up. The body mass index was calculated as they worked, and the ways in which they performed the
weight (kg)/height (m)2. tasks (35). Each subject was assigned a specific value of
In addition, a self-administered questionnaire was used noise exposure on the basis of the 8-hour time-weighted
to collect potential confounders in 1998 and 2008. These average equivalent sound level measured in his workplace.
Am J Epidemiol. 2013;177(8):818–825
820 Chang et al.
Exposure Groups
Total Subjects (n = 578)
Characteristics High (n = 152) Intermediate (n = 221) Low (n = 205) P Value
Mean (SD) No. % Mean (SD) No. % Mean (SD) No. % Mean (SD) No. %
Age at entry, 27.6 (4.6) 27.5 (5.4) 28.0 (5.6) 27.7 (5.3) 0.504a
years
Employment 10.2 (5.0)b 9.0 (4.8)c 10.9 (6.0) 9.8 (5.2) 0.003a
duration,
years
Body mass 24.0 (3.0) 23.8 (3.1) 23.6 (2.9) 23.8 (3.0) 0.493a
indexd
HDL, mg/dl 45.4 (6.9) 46.3 (10.6) 46.0 (8.3) 46.0 (8.9) 0.909a
LDL, mg/dl 115.1 (30.4) 114.5 (30.3) 112.3 (28.8) 113.8 (29.8) 0.609a
Abbreviations: HDL, high-density lipoprotein; LDL, low-density lipoprotein; SD, standard deviation.
a
Kruskal-Wallis test of the difference among the 3 groups.
b
Mann-Whitney test for a significant difference (P < 0.05) compared with the intermediate-exposure group.
c
Mann-Whitney test for a significant difference (P < 0.05) compared with the low-exposure group.
d
Body mass index: weight (kg)/height (m)2.
χ test for a significant difference (P < 0.05) compared with the low-exposure group.
e 2
χ test for a significant difference (P < 0.05) compared with the intermediate-exposure group.
g 2
To avoid an exposure bias due to the use of hearing- earplugs and an average of 62% for earmuffs), a comfort
protective devices at work, we calculated each participant’s factor of 0.5 (because of the minimal percentage of comfort
level of noise reduction according to the noise reduction related to compliance with usage of hearing-protective
rating of the hearing-protective devices that he wore (29 dB devices) (36), and the percentage of working time that he
for earplugs and 25 dB for earmuffs), the protection levels used the hearing-protective devices (36, 37). We used the
of the hearing-protective devices (an average of 28% for hearing-protective device-adjusted value of the 8-hour
Am J Epidemiol. 2013;177(8):818–825
Occupational Noise and Hypertension 821
Table 2. Occupational Noise Exposure at Baseline in 3 Study Groups, Taichung, Taiwan, 1998–2008
time-weighted average equivalent sound level to classify We used a manual stepwise regression to build the final
participants into 3 exposure groups by selecting the median model because only one variable of body mass index
and third quartile as cutoff points in the distribution of (P = 0.005) was retained in the final step using an automatic
noise exposure among all subjects. The 578 workers were stepwise procedure. A basic model was examined first that
subdivided into a high-exposure group (n = 152; noise included age at baseline for biological plausibility and 2
level: ≥85 dBA), an intermediate-exposure group (n = 221; dummy variables of exposure categories. The basic model
noise level: 80–<85 dBA), and a low-exposure group was then enlarged to include 2 variables (i.e., body mass
(n = 205; noise level: <80 dBA). index and employment duration) that were significantly
associated with the risk of hypertension in the simple Cox
regression models. The final model included all variables in
Statistical analysis
the extended model and important risk factors of hyperten-
We first used the Shapiro-Wilk test to determine the nor- sion reported in the previous literature (32–33, 40), includ-
mality of continuous variables. The Kruskal-Wallis test was ing socioeconomic status (using educational levels or job
then used to perform multiple comparisons of continuous positions as surrogates), cigarette use, alcohol intake, and
variables among the 3 groups for the nonnormal distribution, regular exercise. In the sensitivity analysis, the 2 variables
and a 1-way analysis of variance was used for the same com- of salt intake and family history of hypertension that had
parison of continuous variables that were normally distribu- been collected only in 2008 were included in the model.
ted. We also used the χ2 test to compare the difference in The SAS standard package for Windows, version 9.2 (SAS
dichotomous variables among the 3 groups. For those Institute, Inc., Cary, North Carolina), was used for the sta-
groups with significant differences, the Mann-Whitney test tistical analyses. The significance level was set at 0.050 for
(or t test) and the χ2 test were used to compare the high- and all tests.
intermediate-exposure groups with the low-exposure group
for continuous and dichotomous variables. RESULTS
To compare individual differences in SBP and DBP
between the baseline and follow-up measurements, we used Table 1 summarizes the demographic characteristics of
the linear mixed-effect regression models for each exposure 578 participants at baseline. Significant differences were
group (38, 39). The fixed effects in the mixed model identified among the exposure groups in the mean value of
included all variables in the final model at baseline and the employment duration, the number of individuals with an
use of antihypertension medication in 2008. Individual sub- educational level of ≤12 years, and whether individuals
jects were used as a random effect. We used the first-order used hearing-protective devices at work. Workers in the
autoregressive model for covariance structures because of high- and intermediate-exposure groups were more likely to
the minimizing value of Akaike’s Information Criterion in have an educational level of ≤12 years and were less likely
both SBP and DBP measurements (38, 39). to never use hearing-protective devices at work than those
To avoid the information bias in observed person-years in the low-exposure group. In addition, high-exposure
due to the availability of only 1 blood pressure measurement workers had a significantly higher mean of employment
within the past 10 years, we identified hypertensive cases duration and were less likely to never use hearing-protective
by questionnaire, blood pressure measurements, and total devices at work compared with the intermediate-exposure
hypertensive cases used as the outcomes to perform the regres- workers. In contrast, intermediate-exposure workers had a
sion analyses. We used Cox proportional hazard regressions significantly lower mean of employment duration than did
and calculated relative risks with 95% confidence intervals the low-exposure workers.
to compare the differences in incidence of hypertension Table 2 shows the measurements of occupational noise
among groups while controlling for potential confounders. exposure for the 3 groups. Significant differences were
Am J Epidemiol. 2013;177(8):818–825
822 Chang et al.
Linear mixed-effect regression models were used to test the difference between the baseline and follow-up measurements after controlling for age at baseline, antihypertension
identified in the mean noise levels among the 3 groups
P Value
0.479b
either before or after an adjustment for hearing-protective
device use. The high- and intermediate-exposure groups
Baseline–Follow-up
were exposed to significantly higher mean values of noise
Differencea
−0.6, 3.2
−0.3, 3.2
0.1, 4.8
compared with the low-exposure group both before and
95% CI
after an adjustment for hearing-protective device use.
Table 3 shows the changes in blood pressure between
the baseline and follow-up measurements in the 3 groups. Al-
Mean
1.3
though there were no significant differences in SBP and DBP
1.4
2.5
among these 3 groups at baseline or follow-up, the SBP of
both the high-exposure (P = 0.035) and the intermediate-
P Value
0.194b
DBP, mm Hg
81.6 (8.7)
82.5 (8.8)
83.3 (8.6)
0.260b
medication, body mass index, employment duration, educational level, cigarette use, alcohol intake, and regular exercise.
Mean (SD)
0.874b
−1.0, 4.0
2.5, 7.3
0.2, 6.2
Mean
exposure group.
In sensitivity analyses, the association between occupa-
P Value
0.597b
hypertension (P = 0.656).
123.1 (12.1)
122.0 (11.7)
122.9 (11.8)
Mean (SD)
DISCUSSION
Am J Epidemiol. 2013;177(8):818–825
Occupational Noise and Hypertension 823
Table 4. Associations Between Different Noise Exposure Levels and Incidence of Hypertension, Taichung, Taiwan, 1998–2008
Abbreviations: ARR, adjusted relative risk; CI, confidence interval; dBA, A-weighted decibel(s); RR, relative risk.
a
The Cox proportional hazards regression adjusted for biological plausibility (i.e., age at baseline) as the basic model.
b
The Cox proportional hazards regression adjusted for age at baseline and significant factors in simple Cox regression models (such as
body mass index and employment duration) as the extended model.
c
The Cox proportional hazards regression adjusted for all variables in model 1, model 2, and important risk factors reported in previous
literature (i.e., educational level, cigarette use, alcohol intake, and regular exercise) as the final model.
d
Subjects reported that a physician had previously given them a diagnosis of hypertension.
e
Subjects had a mean value of resting systolic blood pressure of ≥140 mm Hg or a mean value of resting diastolic blood pressure of
≥90 mm Hg.
occupational noise exposure. Moreover, the present study In addition, the inverse-U–shaped dose-response rela-
overcame the lack of individual risk factors of hypertension tionship for measured hypertension indicated a role of
in the previous study (10) to provide strong evidence that exposure duration in the etiology of the disease. Although
hypertension was associated with occupational noise expo- we used Cox regressions to account for varying time of ex-
sure. We suggest that the currently regulated threshold for oc- posure, the magnitudes of effects on measured hyperten-
cupational noise exposure may expand the prevention of sion and diagnosed hypertension in the high-exposure
noise-induced hearing loss (41) to the risk of hypertension, group were similar (i.e., around 2-fold), yet the effect
which is a leading cause of cardiovascular diseases. halved from 2.22 to 1.16 in the intermediate-exposure
We also observed an exposure-response pattern between group. These findings may indicate a “plateau” effect of ex-
noise-exposure levels and the risk of hypertension. Our posure intensity that is independent of exposure duration.
results were concordant with the findings in a cross-sectional Therefore, when the exposure intensity is not the highest,
study showing that increasing noise exposure from 75 to the exposure duration is likely the second component of ex-
104 dBA was associated with an increasing prevalence of posure that “kicks in” the dysregulation of the nervous and
hypertension in female textile mill workers (16). One hormonal systems, leading to the disease.
cohort study reported a significantly increasing risk of The significant increases in SBP between the baseline
hypertension with cumulative noise exposure ranging from and follow-up measurements were also observed among
95 dBA × years to >115 dBA × years among male workers workers exposed to 80–<85 dBA and those exposed
(10). The same exposure metrics (i.e., 4 categories from to ≥85 dBA. These findings were consistent with the
<85 dBA × years to ≥95 dBA × years) were applied to our results in a cohort study showing that male workers
analyses, but no significant dose-response relationship exposed to ≥85 dBA and using hearing-protective devices
(adjusted RR = 1.03, 95% CI: 0.87, 1.21) was found in the had a mean increase of 3.8 mm Hg in SBP over 9 years of
Poisson regression. The possible reason for different expo- follow-up, which is a significantly higher increase than that
sure conditions resulting in the same dose-response associ- observed in office workers (11). Additionally, workers
ation might be that workers did not change jobs and tasks exposed to ≥85 dBA exhibited a significant increase in
in present and previous studies (16) and that there was turn- DBP between the baseline and follow-up measurements as
over in another cohort (10). well as a significantly higher DBP at follow-up compared
Am J Epidemiol. 2013;177(8):818–825
824 Chang et al.
with those exposed to <80 dBA. These findings provided the greatest changes in lifestyle. However, this reason
the evidence to explain the association between occupation- seems improbable because the prevalence rates of risk
al noise exposure above 85 dBA and the higher risk of factors related to hypertension, such as cigarette use (58%)
hypertension. and alcohol intake (63%), declined significantly (27% and
The strengths of the present study include a cohort-study 21%, respectively; both P’s < 0.001) in the high-exposure
design, detailed assessments of personal exposure histories, group at the follow-up examination. In addition, blood
and comprehensive controls for most potential confounders pressure was measured twice over the past 10 years. This is
of hypertension. In addition, occupational noise exposure our first attempt to longitudinally observe the occupational
adjusted for the use of hearing-protective devices may noise exposure and the incidence of hypertension among
avoid the misclassification of study subjects and an over- these workers. The 4-year period of follow-up will continue
estimation of noise exposure to produce the consistent results in 2012 to circumvent limitations imposed by lifestyle
associated with road traffic noise exposure in environmental changes over time and to have additional measurements of
epidemiologic studies (34, 42). noise exposure and blood pressure.
Am J Epidemiol. 2013;177(8):818–825
Occupational Noise and Hypertension 825
4. Selander J, Nilsson ME, Bluhm G, et al. Long-term exposure 24. Talbott E, Helmkamp J, Matthews K, et al. Occupational
to road traffic noise and myocardial infarction. Epidemiology. noise exposure, noise-induced hearing loss, and the
2009;20(2):272–279. epidemiology of high blood pressure. Am J Epidemiol.
5. Willich SN, Wegscheider K, Stallmann M, et al. Noise 1985;121(4):501–514.
burden and the risk of myocardial infarction. Eur Heart J. 25. van Dijk FJ, Souman AM, de Vries FF. Non-auditory effects
2006;27(3):276–282. of noise in industry. VI. A final field study in industry. Int
6. Virkkunen H, Harma M, Kauppinen T, et al. The triad of Arch Occup Environ Health. 1987;59(2):133–145.
shift work, occupational noise, and physical workload and 26. Hirai A, Takata M, Mikawa M, et al. Prolonged exposure to
risk of coronary heart disease. Occup Environ Med. 2006; industrial noise causes hearing loss but not high blood
63(6):378–386. pressure: a study of 2124 factory laborers in Japan.
7. Virkkunen H, Kauppinen T, Tenkanen L. Long-term effect of J Hypertens. 1991;9(11):1069–1073.
occupational noise on the risk of coronary heart disease. 27. Hessel PA, Sluis-Cremer GK. Occupational noise exposure
Scand J Work Environ Health. 2005;31(4):291–299. and blood pressure: longitudinal and cross-sectional
8. Sorensen M, Hvidberg M, Andersen ZJ, et al. Road traffic observations in a group of underground miners. Arch Environ
noise and stroke: a prospective cohort study. Eur Heart J. Health. 1994;49(2):128–134.
Am J Epidemiol. 2013;177(8):818–825