Noise Exposure During Pregnancy, Birth Outcomes And Fetal Development: Meta-Analyses Using Quality Effects Model

Abstract

Background: Many women are exposed daily to high levels of occupational and residential noise, so the effect of noise exposure on pregnancy should be considered because noise affects both the fetus and the mother herself. However, there is controversy in the literature regarding the adverse effects of occupational and residential noise on pregnant women and their fetuses.
Aim: The aim of this study was to conduct systematic review of previously analyzed studies, to add additional information omitted in previous reviews and to perform meta-analyses on the effects of noise exposure on pregnancy, birth outcomes and fetal development.
Material and methods: Previous reviews and meta-analyses on the topic were consulted. Additionally, a systematic search in MEDLINE, EMBASE and Internet was carried out. Twenty nine studies were included in the meta-analyses. Quality effects meta-analytical model was applied.
Results: Women exposed to high noise levels (in most of the studies ≥ 80 dB) during pregnancy are at a significantly higher risk for having small-for-gestational-age newborn (RR = 1.19, 95% CI: 1.03, 1.38), gestational hypertension (RR = 1.27, 95% CI: 1.03, 1.58) and infant with congenital malformations (RR = 1.47, 95% CI: 1.21, 1.79). The effect was not significant for preeclampsia, perinatal death, spontaneous abortion and preterm birth.
Conclusion: The results are consistent with previous findings regarding a higher risk for small-for-gestational-age. They also highlight the significance of residential and occupational noise exposure for developing gestational hypertension and especially congenital malformations.

Full text

204
Folia Medica 2014; 56(3): 204-214
Copyright © 2014 Medical University, Plovdiv
doi: 10.2478/folmed-2014-0030
NOISE EXPOSURE DURING PREGNANCY, BIRTH OUTCOMES AND FETAL
DEVELOPMENT: META-ANALYSES USING QUALITY EFFECTS MODEL
Angel M. Dzhambov1*, Donka D. Dimitrova2, Elena D. Dimitrakova3
1Faculty of Medicine, 2Department of Health Care Management, Health Economics and Primary Care, Faculty
of Public Health, 3Department of Obstetrics and Gynecology, Faculty of Medicine, Medical University, Plovdiv,
Bulgaria
ШУМОВАЯ ЭКСПОЗИЦИЯ ВО ВРЕМЯ БЕРЕМЕННОСТИ, ИСХОДЫ РОДОВ,
ЭМБРИОНАЛЬНОЕ РАЗВИТИЕ; МЕТА-АНАЛИЗЫ, С ПРИМЕНЯЮЩИЕ QUALITY
EFFECTS MODEL
Ангел М. Джамбов1, Донка Д. Димитрова2, Елена Д. Димитракова3
1Медицинский факультет, 2Кафедра мениджмента здравоохрнения, экономики здравоохранения и общей
медицины, Факультет общественного здоровья, 3Кафедра акушерства и гинекологии, Медицинский
факультет, Медицинский университет, Пловдив, Болгария
ABSTRACT
BACKGROUND: Many women are exposed daily to high levels of occupational and residential noise, so the effect
of noise exposure on pregnancy should be considered because noise affects both the fetus and the mother herself.
However, there is a controversy in the literature regarding the adverse effects of occupational and residential
noise on pregnant women and their fetuses. AIM: The aim of this study was to conduct systematic review of previ-
ously analyzed studies, to add additional information omitted in previous reviews and to perform meta-analyses
on the effects of noise exposure on pregnancy, birth outcomes and fetal development. MATERIAL AND METHODS:
Previous reviews and meta-analyses on the topic were consulted. Additionally, a systematic search in MEDLINE,
EMBASE and Internet was carried out. Twenty nine studies were included in the meta-analyses. Quality effects
meta-analytical model was applied. RESULTS: Women exposed to high noise levels (in most of the studies ≥
80 dB) during pregnancy are at a signifi cantly higher risk for having small-for-gestational-age newborn (RR =
1.19, 95% CI: 1.03, 1.38), gestational hypertension (RR = 1.27, 95% CI: 1.03, 1.58) and infant with congenital
malformations (RR = 1.47, 95% CI: 1.21, 1.79). The effect was not signifi cant for preeclampsia, perinatal death,
spontaneous abortion and preterm birth. CONCLUSION: The results are consistent with previous fi ndings regarding
a higher risk for small-for-gestational-age. They also highlight the signifi cance of residential and occupational
noise exposure for developing gestational hypertension and especially congenital malformations.
Key words: noise exposure, pregnancy, birth outcomes, fetal development, meta-analysis, quality effects model
Folia Medica 2014; 56(3): 204-214
Copyright © 2014 Medical University, Plovdiv
РЕЗЮМЕ
ВВЕДЕНИЕ: Так как многие женщины живут и работают в условиях шума, то влияние шумовой экспозиции
на беременность следует принимать во внимание, так как шум имеет прямые и косвенные эффекты
на плод и мать. Несмотря на это в литературе существует противоречивое мнение относительно
нежелательных эффектов шума и дома, и на работе на беременных женщин и их плод. ЦЕЛЬ: Работа
ставит себе целью провести систематический обзор уже анализированных исследований, добавить
дополнительную информацию, невключенную до сих пор в имеющиеся обзоры и сделать мета-анализы
эффектов шумовой экспозиции на беременность, исход родов и развитие плода. МАТЕРИАЛ И МЕТОДЫ:
Авторы сделали справку со существующими обзорами на эту тему. Кроме того проведен систематический
поиск в MEDLINE, EMBASE и Internet. 29 исследований включено в мета-анализы; применена quality ef-
fects мета-аналитическая модель. РЕЗУЛЬТАТЫ: Женщины, экспонированные на высокие шумовые уровни
(в большинстве исследований ≥ 80 dB), во время беременности подвергнуты значительно более высокому
риску родов ребенка небольшой массы тела для гестационного возраста (RR = 1.19, 95% CI: 1.03, 1.38),
риску гестационной гипертонии (RR = 1.27, 95% CI: 1.03, 1.58) и риску родов ребенка с врожденными
мальформациями (RR = 1.47, 95% CI: 1.21, 1.79). Эффект не оказался значимым для преэклампсии,
перинатальной смерти, спонтанного аборта и преждевременных родов. ЗАКЛЮЧЕНИЕ: Результаты
подтверждают уже установленный повышенный риск родов маленького для гестационного возраста
ребенка; подчеркивают также значимость комунальной и трудовой шумовой экспозиции для развития
гестационной гипертонии и особенно для развития врожденных мальформаций.
Ключевые слова: шумовая экспозиция, исход беременности, развитие эмбриона, мета-анализы, quality
effects model Folia Medica 2014; 56(3): 204-214
© 2014 Все права защищены. Медицинский университет, Пловдив
Public Health Care
Article’s history: Received: 30 January 2014; Received in a revised form: 22 May 2014 Accepted: 01 June 2014
*Correspondence and reprint request to: A. Dzhambov, Faculty of Medicine, Medical University, Plovdiv;
E-mail: angelleloti@gmail.com; Mob.: 359 89 79 50 802
15A Vassil Aprilov St., 4002 Plovdiv, Bulgaria

205
Folia Medica 2014; 56(3): 204-214
© 2014 Medical University, Plovdiv
Noise Exposure During Pregnancy, Birth Outcomes and Fetal Development:
Meta-Analyses Using Quality Effects Model
INTRODUCTION
Noise pollution is continuing to grow as a result of
urbanization and industrialization.1 Eighty million
people in Europe alone are exposed to community
noise levels above 65 dB,2 and it has been estimated
that up to 2% of the gross domestic product as well
as one million healthy life years are lost every year
in Europe due to noise pollution3,4. Occupational
noise is also associated with adverse health outcomes:
hearing loss, performance impairment, cardiovascular
diseases and mental exhaustion.5 In 2000, as much
as 20% of all European workers were exposed to
noise so loud that it required raising their voice to
talk to other people.6 There are various risk groups
particularly sensitive to environmental noise and
one of those is pregnant women.
As a unique environmental stressor noise exerts
both direct and indirect effects on humans. Noise
higher than 85 dB might lead to permanent hearing
impairment.7 Because of its indirect effects noise is
considered a risk factor for hypertension, ischemic
heart disease, annoyance and sleep disturbance.8-12
Approximately 210 000 of all coronary incidents
in Europe every year can be attributed to traffi c
noise exposure.13
As many women live and work in highly noisy
environments, noise exposure is in the focus of
attention of public health experts when pregnancy
is involved. Generally, some authors find that
women might have elevated risk of noise induced
cardiovascular alterations.14,15 Others even pro-
posed that the higher noise sensitivity of women
should be considered an independent risk factor
for cardiovascular diseases.16 If changes in cardio-
vascular and endocrine function occur under noise
conditions, they can have some adverse effects
on human pregnancy. The human body reacts to
environmental noise with a general stress response
mechanism leading to neuroendocrine and cardio-
cardiovascular alterations.17 It activates the amygdala,
some cortical limbic and hypothalamic centres18,
thus affecting synaptic links in the reticular forma-
tion and mesencephalon, as well as emotional and
cognitive pathways of perception through cortical
and subcortical structures19,20. Ultimately this leads
to stimulation of the sympathetic-adrenal axis.21
Stress-release of maternal catecholamine may in-
crease blood pressure and uterine reactivity and thus
decrease placental function leading to hypoxia of
the fetus.22 Maternal cortisol might pass through
the placental barrier and interfere in the regulation
of the fetal hypothalamic-pituitary-adrenal axis, or
stimulate the placenta to secrete corticotropin releas-
ing hormone.23 On other hand, as the intrauterine
environment is rich with externally generated noise
passing through the abdominal wall24, noise energy
might affect the fetus directly25,26. Sound is easily
transmitted to the fetus through the abdominal wall
which has an attenuation of about 10 dB or less
in the low frequency range.
However, there is a controversy in the literature
regarding the adverse effects of both occupational
and residential noise on pregnant women and their
fetuses. Some authors fi nd association with birth
defects, shortened gestation and decreased birth
weight, while others fi nd no effect.27 Recently a
systematic review by Hohmann et al.28 found no
evidence for effect of noise exposure on pregnancy,
birth outcomes or fetal development. It, however,
did not review all evidence that was present at that
moment. Previously the government of Quebec had
reported a thorough and extensive meta-analysis on
the effects of workplace noise on pregnancy and
found evidence suggesting that in the presence of
occupational noise exposure ≥ 85 dB pregnancy
leads to low birth weight for the gestational age, and
might possibly cause spontaneous abortion, preterm
birth, preeclampsia and gestational hypertension.29
Both these papers have their limitations: Hohmann
et al.28 applied only qualitative appraisal of the data
and did not include all available evidence, while
Croteau, et al.29 reviewed only studies dealing with
occupational noise. Although they adopted sophis-
ticated quality assessment methodology and used
meta-regressions, they adhered to the DerSimonian-
Laird random effects meta-analytical model, which
has been criticized for its limitations.30,31 Finally,
since the systematic literature searches of these
two studies were performed, new evidence on the
topic might have been accumulated.
AIM
In this paper we focus only on the effect of in-
trauterine noise exposure on pregnancy, birth out-
comes and fetal development. The aim of this study
was to perform a systematic review of previously
analyzed studies28,29, to add additional information
not included in previous research and to perform
meta-analyses on the effects of noise exposure on
pregnancy, birth outcomes and fetal development.
MATERIAL AND METHODS
SEARCH STRATEGY
This research was greatly facilitated by previously
published high quality papers on the topic.28,29 Most

206
A. Dzhambov et al
Folia Medica 2014; 56(3): 204-214
© 2014 Medical University, Plovdiv
of the literature sources on the relationship noise
– pregnancy outcomes was readily collected, sum-
marized, reviewed and available from government
of Quebec’s report.29 Additionally, we carried out
systematic search in MEDLINE, EMBASE and gen-
erally the Internet using the search engines PubMed,
ScienceDirect and Google (studies published before
December 23, 2013). Two of the authors performed
the searches independently. English, Russian and
Spanish language articles were screened on three
levels: titles, abstracts and full-texts. Inclusion cri-
teria were: studies dealing with noise exposure as a
primary or secondary factor and its effect on preg-
nancy, intrauterine development or birth outcomes;
observational and experimental studies involving
humans or reviews/meta-analyses. Papers describing
research on hearing impairment due to fetal noise
exposure or research in intensive care units were
excluded as they were not in our scope. The fol-
lowing free-term keyword combinations were used:
“noise + pregnancy”, “ruido + embarazo” and “шум
+ беременность”. PubMed retrieved 653 results with
no fi lters applied while in ScienceDirect English
language search was limited to topics associated
with prenatal development, pregnancy outcomes and
intrauterine imaging diagnostics – 621 results were
retrieved in English, 278 in Spanish and none in
Russian. Additionally, Google search provided two
more papers in Spanish. Hand searching through the
reference list of a report of the American Academy
of Pediatrics32 provided fi ve articles not included
in previous reviews33-37. Hand search of the refer-
ence lists of articles available at the web-page of
Scandinavian Journal of Work, Environment &
Health retrieved fi ve more studies previously not
meta-analyzed.38-42 The abstract of Knipschild et
al.’s paper43 was retrieved and additional data about
it had been abstracted by Hohmann et al.28
Studies which, after careful consideration, were
deemed irrelevant to the topic were dropped from
further processing (for example, Heidam40). Due to
limited access to repository information, some of
the articles could not be retrieved in full-text and
if the abstract or previous reviews did not provide
suffi cient data to assess their quality or calculate
effect size estimates, they were not included in
the meta-analyses.38,39 Inter-rater agreement for the
included studies was acceptable (Kappa = 0.74) and
any discrepancies were resolved after discussion
with a third expert.
QUALITY ASSESSMENT
Individual study quality assessment methodology was
adopted from Croteau et al.29 Briefl y, their checklist
assigned a total of 18 points for perfectly credible
study for external and internal validity according to
the following elements: country where the study was
conducted, period when the study was conducted,
mode of selection of the population, participation
rate, defi nition of the effect on pregnancy, mea-
surement of the effect on pregnancy, defi nition of
noise exposure, selecting the comparison group,
measurements and covariates. For more detailed
information the reader is referred to Croteau, et
al.29 Individual study scores were recalculated as
relative shares from the maximum 18 points and
re-scaled between 0 for a low quality study and 1
for a perfectly scored.30,31
DATA EXTRACTION AND PROCESSING
Meta-analyses were performed to estimate the
effects of noise exposure (both occupational and
residential) during pregnancy on the risk for ges-
tational hypertension (blood pressure > 140/90 mm
Hg), fetal malformations, small-for-gestational-age
infants (≤ 2500 g/smaller than the 10th percentile),
preeclampsia, preterm birth (≤ 37 gestational week)
and perinatal death. Two of the studies had low
birth weight as outcome43,44, while the rest assessed
small-for-gestational-age infants according to the
classifi cation of Croteau et al. Although these are
different entities, Croteau et al. used the results of
Peoples-Sheps et al.45 as representative of small-
for-gestational-age, while in fact they measured low
birth weight (≤ 2500 g). They used low birth weight
as a proxy for small-for-gestational-age claiming
that it could be the result of preterm birth, small
for gestational age, or both. Magann et al.46 on the
other hand, assessed intrauterine growth restriction,
but this outcome was also interpreted as small for
gestational age by Croteau, et al29. Hence, we felt
comfortable including the two studies mentioned
above44,43 along with the others representing small-
for-gestational-age infants.
MetaXL v.1.4 add-in for Excel (http://www.
epigear.com) was used to conduct the analyses.
Additional data imputation was performed with
Excel spreadsheets or by hand. Relative risks (RR)
were chosen as effect size estimates.
Twenty nine studies were included in the meta-
analyses. Initial data of about 24 of the studies43-67
were extracted from Croteau et al29. For the studies
that we retrieved in full text the effect size esti-
mates reported by Croteau et al. were confi rmed.
However, careful read of their report reveals that
they actually mixed odds ratio (OR), RR and in

207
Folia Medica 2014; 56(3): 204-214
© 2014 Medical University, Plovdiv
Noise Exposure During Pregnancy, Birth Outcomes and Fetal Development:
Meta-Analyses Using Quality Effects Model
some occasions expected to observed ratio. Inter-
preting OR as RR leads to overestimation of the
effect. Moreover, when a straightforward estima-
tion of RR is an option given the original study
data, there is no reason to use OR as a substitute
to RR.68 On the other hand, using OR in this
case might be justifi ed because it is adjusted for
relevant confounders like demographics, previous
pregnancies, socio-economic class, other exposures,
etc. Therefore, whenever possible we calculated the
adjusted RR from the original data in the studies
according to Zhang & Yu69, and when it was not,
we adhered to the adjusted OR. For the fi ve stud-
ies that we identifi ed and have not been previously
meta-analyzed we calculated RR.33,41-43,70 Studies
with zero events in control group were included
in the meta-analyses in order to provide a more
conservative estimate of the effect size.49 In this
case the standard continuity correction of 0.5 was
used according to Cox, 1970 (cited by Friedrich et
al.71). When RR or OR were reported for several
noise exposure categories, those using the high-
est exposure as a cut point were included in the
meta-analyses.
All effect size estimates for the different out-
comes were entered into MetaXL which was set
to produce pooled RR. Of all data reported in the
studies we extracted the effect sizes adjusted for
most of the relevant confounders. The approach of
Zhang & Yu69 is criticized72 and it produces too
narrow 95% confi dence intervals, but given that
most outcome measures in these meta-analyses
are relatively rare in the population and that in
environmental epidemiology we are often inter-
ested in the pooled point estimate rather than the
95% confi dence intervals around it, we did not
expect that our fi ndings would be associated with
signifi cant imprecision.
Quality effects meta-analytical model was applied.
According to Overton73 if the studies included in
the meta-analysis differ in systematic way from
the possible range in the population, they are not
representative of it and the random effects model
does not apply. Therefore, the quality effects model
proposed by Doi and Thalib was applied.30,31 It
uses a quality index Qi representing the probability
the judgment of that study is credible. From Qi
a study specifi c composite is generated that takes
into consideration study specifi c information and
its relationship to other studies to re-distribute
inverse variance weights.
Heterogeneity was explored using the chi-square
test. The quantity of heterogeneity across studies
was measured by the I2 statistic.74 According to
the I2 values, heterogeneity was considered low
(< 25%), moderate (25 – 50%) and high (> 50%).
Sensitivity analysis was performed by assessing
the contribution of individual studies to the sum-
mary effect estimate by excluding each trial, one
at a time, and computing meta-analysis estimates
for the remaining studies. Publication bias was
not assessed due to inter-study heterogeneity and
the limited number of studies included for each
individual outcome.75 Results were considered
statistically signifi cant at p < 0.05. All analyses
were carried out with MS Excel v. 2010.
RESULTS
In this paper we will fi rst present a brief overview
of those studies previously not subject to systematic
review.33,41,42,67 Table 1 reports their characteristics
according to the quality assessment procedure of
Croteau, et al.29 For additional information regarding
the rest of the studies included in the meta-analyses
the reader is kindly asked to consult Croteau, et
al.29 and Hohmann et al.28
Of all included studies nine were case-con-
trol41,52,53,57-60,65,67, two were cross-sectional43,70,
four were cohort46,47,49,64, 13 were retrospec-
tive42,44,45,48,50,51,54-56,61-63,66 and one was ecological33.
(Ecological study was defi ned as one using aggre-
gated data.) The sample sizes varied considerably
from 179 to 225 146. Small-for-gestational-age
was an outcome in 12 studies, preterm birth in
11, gestational hypertension in seven, spontaneous
abortion in fi ve, preeclampsia in three (Irwin et al.62
was included twice with results for nulliparous and
multiparous women), three assessed perinatal death
and fi ve - congenital malformations. Ten studies
assessed more than one outcome. Adjustments for
relevant confounders varied but they were to some
extent accounted for by the quality scores. Finally,
two studies were dealing with residential rather
than occupational noise exposure.
There is a 19% risk for small-for-gestational-age
if the mother has been exposed to ≥ 80 dB during
pregnancy (Fig. 1). All studies used a cut point of
approximately 80 dB risk noise exposure assessed
either by specifi c question about the acoustic envi-
ronment at work or by quantifi cation by industrial
hygienists. Sensitivity analysis revealed that by
excluding each study one at a time the pooled RR
remained signifi cant in the range 1.16 – 1.27, with
the only exception of excluding McDonald et al.44
(RR = 1.15, 95% CI: 0.97, 1.36). However, most
of the studies used OR and this was one of the

208
A. Dzhambov et al
Folia Medica 2014; 56(3): 204-214
© 2014 Medical University, Plovdiv
Table 1. Characteristics of the studies not previously reviewed or meta-analyzed (description is given according to Croteau et al.29)
Study Design External
validity
(2) Population Effects on pregnancy Noise exposure Controlling for confound-
ers
Method of
selection
(2)
Response
rate
(2)
Defi nition
(1)
Measure
(1)
Defi nition
(1)
Control
group
(1)
Measure
(4)
Personal
factors
(2)
Other
exposures
(2)
Nurminen,
198942 Retrospective
study,
n = 1 475
Finland,
1976 –
1982, work-
ers
(2)
Mothers of
infants with
malforma-
tions
(2)
≥ 85%
(2) Gestational
hypertension
(≥ 20 mmHg
increase)
(1)
Offi cial
medical
fi les
(1)
Laeq (8 h)
≥ 80 dB
(1)
Laeq (8 h) <
80 dB
(1)
Industrial
hygienists as-
sessment and
self-report
job descrip-
tion
(4)
Age ≥ 35
yrs, previous
pregnancies,
smoking, al-
cohol, drugs/
infection in
1st trimester
(2)
Solvent
exposure,
physical
load
(2)
Bendok-
iene et al,
201170
Cross-sec-
tional,
n = 3 121
Lithuania,
2007 – 2009
(2)
Pregnant
women from
prenatal care
practices
(2)
79%
(0.5) Gestational
hypertension
(≥140 or ≥ 90
mm/Hg)
(1)
Measured
by physi-
cian
(1)
LAeq (24 h)
≥ 61 dB
(1)
LAeq (24 h)
< 61 dB
(1)
Noise map
data
(4)
Demograph-
ics, diseases,
social status,
marital status,
BMI, alcohol,
smoking (1.5)
None
(0)
Jones &
Tauscher,
197833
Ecological
(aggregated
data),
n = 225 146
North
America,
1970 – 1972
(2)
Births in the
population
close to an
airport
(2)
All
registered
births
(2)
Malformations
(excluding
polydactylia)
(1)
Offi cial
medical
records
(1)
≥ 90 dB
(1) < 90 dB
(1) Noise map
contour
(4)
None
(0) None
(0)
Kurppa et
al, 198341 Case-control,
preliminary
results,
n = 2 094
Finland,
1976 – 1978
(2)
Mothers
exposed to
occupational
chemicals
and noise
pollution
during preg-
nancy (2)
95%
(2) Malformations
(1) Offi cial
medical
records
(1)
≥ 82 dB
(1) < 82 dB
(1) Self-report
exposure and
hygienist as-
sessment
(4)
Matched
case-control
pairs
(2)
Solvents,
pesticides,
radiation,
metals, dis-
infectants
(2)

209
Folia Medica 2014; 56(3): 204-214
© 2014 Medical University, Plovdiv
Noise Exposure During Pregnancy, Birth Outcomes and Fetal Development:
Meta-Analyses Using Quality Effects Model
two studies for which we calculated unadjusted
RR, which might explain the discrepancies. The
main source of heterogeneity was McDonald et
al.55 As Croteau et al.29 point out, the studies of
McDonald et al. compared noise exposed workers
to the general population rather than to non-exposed
workers, which makes these studies a bit problem-
atic. Moreover, this 19% risk should be interpreted
with caution because of the overestimation of the
effect that occurs with OR.
In reference to preterm birth, the risk was not
signifi cantly higher for women exposed to ≥ 80 dB
(Fig. 2). Pooled RR remained non-signifi cant (1.03
– 1.07, p > 0.05) after each study was excluded
and the main sources of heterogeneity were the
studies of Peoples-Sheps et al.45 and Luke et al.57
Statistically signifi cant risk for perinatal death
was not associated with noise exposure (Fig. 3).
It varied considerably remaining non-signifi cant
(0.95 – 1.72, p > 0.05), and according to sensitiv-
ity analysis the positive effect was mostly due to
Hartikainen et al.’s results.49
Pooled RR for spontaneous abortion did not
reach statistical signifi cance either (1.05 – 1.33,
p > 0.05) (Fig. 4) and all heterogeneity was due
to Hansteen at al.65 as its exclusion resulted in I2
= 0.00.
For preeclampsia the variability after excluding
individual studies was also considerable (0.93 –
1.23, p > 0.05) but signifi cant pooled RR could
not be produced (Fig. 5).
Mothers exposed to high noise during pregnancy
had an elevated risk (27%) for gestational hyperten-
sion (Fig. 6). That effect became non-signifi cant if
Note. RR – Relative risk; Q, p and I2 –heterogeneity statistics
Figure 1. Forest plot on studies assessing the risk for small-for-gestational-age.
Note. RR – Relative risk; Q, p and I2 –heterogeneity statistics
Figure 2. Forest plot on studies assessing the risk for preterm birth.

210
A. Dzhambov et al
Folia Medica 2014; 56(3): 204-214
© 2014 Medical University, Plovdiv
the studies of Nurminen,42 Nurminen & Kurppa50,
Bendokiene et al.70 or Saurel-Cubizolles et al.61 were
excluded. Nevertheless, we are inclined to accept
that noise exposure during pregnancy is associated
with higher risk of gestational hypertension, not
only because for this outcome the majority of the
estimates were either reported as adjusted RR or
calculated as RR, thus being conservative, but also
because there is strong evidence for the association
of environmental noise with hypertension among
the general population.8,76
Finally, we found 47% increased risk for fetal
malformations in relation to intrauterine noise
exposure (Fig. 7). These studies were completely
homogenous and the pooled RR did not change
when sensitivity analysis was conducted (1.41 –
1.53, p < 0.05).
As this is a considerable risk, we took a closer
look at this outcome in particular. Out of fi ve stud-
ies assessing this outcome, two were reviewed by
Note. RR – Relative risk; Q, p and I2 –heterogeneity statistics
Figure 3. Forest plot on studies assessing the risk for perinatal death.
Note. RR – Relative risk; Q, p and I2 –heterogeneity statistics
Figure 4. Forest plot on studies assessing the risk for spontaneous abortion.
Note. RR – Relative risk; Q, p and I2 –heterogeneity statistics
Figure 5. Forest plot on studies assessing the risk for preeclampsia.

211
Folia Medica 2014; 56(3): 204-214
© 2014 Medical University, Plovdiv
Noise Exposure During Pregnancy, Birth Outcomes and Fetal Development:
Meta-Analyses Using Quality Effects Model
Croteau et al.29 and three were added by us. The
study of Jones & Tauscher33 which had highest
weight was ecological using aggregated data to
compare the incidence of congenital abnormalities
in two populations – one living close to an airport
and exposed to ≥ 90 dB and another one living in
the city centre and exposed to lower noise levels.
This study had the highest number of cases (n =
225 146) in our meta-analyses, but due to lacking
covariates and the problems associated with ag-
gregated data in received 14 out of 18 points for
credibility. Moreover, as the authors noted, living
close to an airport is associated with various air
pollutants, which might be partly responsible for the
increased incidence of malformations. Nevertheless,
excluding this study controversially raised RR to
50%. The same applies to the other high-weighted
study included by us whose exclusion raised the
risk to 53%.41 Kurppa et al.’s study was assigned
18 out of 18 points for credibility.41 These fi ndings
lead us to suggest that calculating RR for these
studies is suffi ciently conservative as we were
mainly concerned with the specifi city of the pooled
RR. Moreover, the fact that we were not able to
compute RR for Zhang et al.51 and Kurppa et al.67
should not affect the results because congenital
malformation are very rare in human populations
(much lower than 10%).
DISCUSSION
KEY FINDINGS
These meta-analyses showed that intrauterine noise
exposure is signifi cantly associated with higher risk
for small-for-gestational-age, gestational hyperten-
sion and congenital abnormalities. In reference to
preterm birth, preeclampsia, perinatal death and
spontaneous abortion the pooled RR did not permit
making such inferences, although for spontane-
ous abortion and preterm birth it was just failing
statistical signifi cance. In comparison to the fi nd-
ings of previous research on the topic, we did not
confi rm the conclusion of no association of noise
exposure with pregnancy or fetal development made
by Hohmann et al.28 The discrepancies might be
caused by the differences in included studies and
by the methods applied – purely qualitative vs.
meta-analysis. On the other hand, the only other
identifi ed meta-analysis on the topic found suffi -
cient evidence of association between occupational
Note. RR – Relative risk; Q, p and I2 –heterogeneity statistics
Figure 6. Forest plot on studies assessing the risk for gestational hypertension.
Note. RR – Relative risk; Q, p and I2 –heterogeneity statistics
Figure 7. Forest plot on studies assessing the risk for congenital malformations.

212
A. Dzhambov et al
Folia Medica 2014; 56(3): 204-214
© 2014 Medical University, Plovdiv
noise exposure during pregnancy and giving birth
to a small-for-gestational-age child (27%, meta-
regression).29 Thus via two different approaches
and research teams this effect was confi rmed. Cro-
teau et al.29 also suspected an effect on gestational
hypertension, spontaneous abortion, preterm birth
and preeclampsia, but did not confi rm it. Probably
the most prominent discordance between fi ndings
is the signifi cant risk for congenital malformations
estimated in our study, while the available data did
not allow Croteau et al.29 to make conclusions about
congenital malformations. However, it is biologi-
cally plausible since the corticosteroids produced in
reaction to noise are toxic to the embryo and since
noise intensity of about 80 dB could increase the
hematoencephalic barrier’s penetrability.77
STRENGTHS AND LIMITATIONS
This study builds on previous systematic research.
We were able to include studies previously not
meta-analyzed or discussed and to extract new
data from studies which were previously reviewed.
Some of these studies reported results for residential
noise, which combined with those dealing with oc-
cupational noise reveal wider picture of this envi-
ronmental pollutant. In addition, adjusted RR were
calculated where possible which deems our results
more conservative and closer to the real effect of
noise exposure on pregnancy. However, human error
should always be considered given that only one of
the authors conducted the statistical tests. Probably
the main strength of these meta-analyses is the use
of quality effects model to estimate the pooled RR.
According to Doi78 the quality effects model should
replace the currently used random effects model as
it always outperforms it. The fact that we did not
include all quality estimates that the study of Cro-
teau et al.29 offered – for example, publication bias,
biological plausibility, precision statistic – should
not be particularly concerning because the quality
effects model requires only that quality scores rank
the analyzed studies with respect to defi ciencies
rather than by fi xed function of bias or being a
function of quantitatively measured bias.78 In other
words, the quality appraisal and scores need not
be refl ecting the actual credibility or limitations of
the studies, as long as they rank them according
to their relative strength of evidence in reference
to each other. Thus the weights are redistributed
away from lesser quality studies. Another issue
with this paper might be mixing together OR and
RR, but as stated, some of the effect sizes could
not be converted to RR. Although our approach for
calculating adjusted RR is criticized72, the arguments
refer mainly to common outcomes. Combining data
from occupational and residential settings might
also raise some concerns, because noise effects
are dependent on both sound intensity and the
frequency spectrum, as well as on the time pattern
of exposure and individuals’ activities which are
disturbed.79 Finally, the limited access to repository
content and the lack of external funding prevented
us from reviewing the full texts of some studies,
so we had to either rely on previously abstracted
data or leave them out of the analyses.
IMPLEMENTATION
The theory that variations in intrauterine environ-
ment might affect fetal development is increasingly
accepted.23 The results of this study might be used
to promote adequate strategies for screening and
prevention of noise-related pregnancy adverse out-
comes. These inferences will have to be replicated
by other meta-analyses and new fi eld studies and,
if found to be adequate, hypertension screening
programs and prenatal diagnostics will have to be
implemented and made priority for women work-
ing or dwelling in highly noisy environments. It
is true that the industrial development and mecha-
nization of all branches of industry have reduced
the number of women working in manufacture, for
example, which might be the cause for the limited
number of recently published studies assessing these
phenomena and their interrelationships. It is also
true that this does not apply with full strength to
poorer countries with limited advance in industrial
technology. Women are undertaking jobs that were
previously “the privilege of men” – bus drivers, train
engineers, heavy machinery operators, etc. – and this
keeps open the question whether they are actually
exposed to lower noise levels in comparison to 50
years ago. Moreover, residential noise is constantly
rising as a result of urbanization, population growth
and modern lifestyle. Factories, traffi c, railway and
aircraft transportation, which are the major sources
of noise in woman’s life, are also producing toxic
gases which have distinct impact on fetal devel-
opment.80 Hence, quantifying the additive effect
of noise exposure will yield clearer picture and
provide guidance for health promotion and disease
prevention in pregnant women, since both noise
and air pollution can affect cardiovascular health
via different pathways.81

213
Folia Medica 2014; 56(3): 204-214
© 2014 Medical University, Plovdiv
Noise Exposure During Pregnancy, Birth Outcomes and Fetal Development:
Meta-Analyses Using Quality Effects Model
CONCLUSIONS
The results confi rm some previous fi ndings re-
garding a higher risk for small-for-gestational-age.
They also highlight the signifi cance of residential
and occupational noise exposure for developing
gestational hypertension and especially congeni-
tal malformations when the mother is exposed to
noise levels ≥ 80 dB. These fi ndings will have to
be confi rmed by new fi eld studies and, if found to
be adequate, hypertension screening programs and
prenatal diagnostics will have to be implemented
and made priority for women working or dwelling
in highly noisy environments.
REFERENCES
1. Lee CSY, Fleming GG. General Health Effects
of Transportation Noise. U.S. Department of
Transportation. Washington, DC; 2002. Retrieved
from: http://www.fra.dot.gov/downloads/RRDev/
health_Final.pdf. Accessed August 6, 2013.
2. European Community. Commission green paper on
future noise policy; 1996. Retrieved from: http://
ec.europa.eu/environment/noise/pdf/com_96_540.
pdf. Accessed: June 3, 2013.
3. WHO. European Centre for Environment and
Health. Burden of disease from environmental noise:
quantifi cation of healthy life years lost in Europe.
Copenhagen: Regional Offi ce for Europe; 2011.
Retrieved from: http://www.euro.who.int/__data/
assets/pdf_fi le/0008/136466/e94888.pdf. Accessed
September 5, 2013.
4. WHO. Prevention of Noise-Induced Hearing Loss.
WHO-PDH Informal Consultation Report, 5, 7, 20;
1997. Retrieved from: http://www.who.int/pbd/deaf-
ness/en/noise.pdf. Accessed September 4, 2013.
5. WHO. Occupational noise: Assessing the burden of
disease from work-related hearing impairment at
national and local levels. Environmental Burden of
Disease Series, No. 9. WHO Document Production
Services, Geneva, Switzerland; 2004. Retrieved
from: http://www.who.int/quantifying_ehimpacts/
publications/en/ebd9.pdf. Accessed December 1,
2013.
6. European Foundation for the Improvement of Liv-
ing and Working Conditions. The Third European
survey on working conditions 2000 (ESWC); 2000.
Retrieved from: http://www.eurofound.europa.eu/
pubdocs/2001/21/en/1/ef0121en.pdf. Accessed Ja-
nuary 15, 2014.
7. Berglund B, Lindvall T, editors. Community Noise.
Archives of the Center for Sensory Research. 1995;
2:1-195. WHO. Retrieved from: http://www.who.
int/docstore/peh/noise/guidelines2.html. Accessed
January 7, 2014.
8. van Kempen E, Babisch W. The quantitative relation-
ship between road traffi c noise and hypertension: a
meta-analysis. J Hypertens 2012;30:1075-86.
9. Sobotova L, Jurkovicova J, Stefanikova Z, et al.
Community response to environmental noise and
the impact on cardiovascular risk score. Sci Total
Environ 2010;408:1264-70.
10. Eriksson C, Rosenlund M, Pershagen G, et al. Air-
craft noise and incidence of hypertension. Epidemi-
ology 2007;18(6):716-21.
11. Öhrström E. Longitudinal surveys on effects of
changes in road traffi c noise annoyance, activity
disturbances, and psycho-social well-being. J Acoust
Soc Am 2004;115:719-29.
12. Hume KI, Brink M, Basner M. Effects of environ-
mental noise on sleep. Noise Health 2012;14:297-
302.
13. Mead MN. Noise Pollution: The sound behind heart ef-
fects. Environ Health Perspect 2007;115(11):A536-7.
14. Jarup L, Babisch W, Houthuijs D, et al. Hypertension
and exposure to noise near airports: the HYENA
study. Environ Health Perspect 2008;116(3):329–33.
15. Bluhm LG, Berglind N, Nordling E, et al. Road
traffi c noise and hypertension. Occup Environ Med
2007;64(2):122-6.
16. Heinonen-Guzejev M. Noise sensitivity – medical,
psychological and genetic aspects [dissertation].
University of Helsinki: Helsinki University Printing
House Helsinki; 2008. Retrieved from: http://www.
doria.fi/bitstream/handle/10024/42979/noisesen.
pdf?sequence=1. Accessed August 15, 2013.
17. Goines L, Hagler L. Noise pollution: a modem
plague. South Med J 2007;100(3):287-94.
18. Ising H, Kruppa B. Health effects caused by noise:
Evidence in the literature from the past 25 years.
Noise Health 2004;6:5-13.
19. Andersson K, Lindvall T. Health effects of commu-
nity noise; evaluation of the Nordic project on “the
health effects of community noise”. Nordic Council
of Ministers, Copenhagen; 1988.
20. Spreng M. Central nervous system activation by
noise. Noise Health 2000;7:49-57.
21. Babisch W. The noise/stress concept, risk assessment
and research needs. Noise Health 2002;4(16):1-11.
22. Wu T-N, Chen L-J, Lai J-S, et al. Prospective study
of noise exposure during pregnancy on birth weight.
Am J Epidemiol 1996;143(8):792-6.
23. de Weerth C, Buitelaar JK. Physiological stress
reactivity in human pregnancy--a review. Neurosci
Biobehav Rev 2005;29(2):295-312.
24. Richards DS, Frentzen B, Gerhardt KJ, et al.
Sound levels in the human uterus. Obstet Gynecol
1992;80:186-90.
25. Gerhardt KJ. Prenatal and perinatal risks of hearing
loss. Semin Perinatol 1990;14:299-304.

214
A. Dzhambov et al
Folia Medica 2014; 56(3): 204-214
© 2014 Medical University, Plovdiv
References truncated due to limitation of refer-
ences’ number for this type of study. All references
are available upon request to the corresponding
author or at: https://docs.google.com/document/
d/1zA0ZHLAaCxLssn9MIBbMNwO_MbpUgA-
4nSaIJ4oYkQ8/edit?usp=sharing.