Exposure to ambient air pollution and prenatal and early childhood health effects potx

Key words: Ambient air pollution, Congenital defects, Intrauterine growth retardation, Low birth weight, Mortality, Preterm delivery Abbreviations: BW = birth weight; CI = confidence inte

DEVELOPMENTAL EPIDEMIOLOGY

Exposure to ambient air pollution and prenatal and early childhood health effects

Marina Lacasan˜a1,2,3, Ana Esplugues1,2 & Ferran Ballester1,4

1

Epidemiology and Statistics Unit, Valencian School of Studies for Health, Valencia, Spain;2University Hospital La Fe, Valencia, Spain;3Center for Environmental Health, Center for Research in Populational Health, National Institute of Public Health, Cuernavaca, Mexico;4Public Health Department, History of Science and Gynaecology, Miguel Herna´ndez University, Alicante, Spain

Accepted in revised form 7 September 2004

Abstract Over the last years, concern for the

pos-sible influence of exposure to air pollutants in

children during gestation or the first years of life

has grown; exposure levels which may be reached

nowadays in our dwellings and in our streets In the

present study evidence over the possible impact of

ambient air pollution on the foetus and the infants

(i.e.: less than 1 year) published during the last

dec-ade, 1994–2003, are revised Studies on infant

mor-tality and exposure to particles show an outstanding

consistence in the magnitude of the effects, despite the

different designs used As a whole, data show that an

increase in 10lg/m3

of particle concentration (mea-sured as PM10) is associated with to about 5%

in-crease in post-neonatal mortality for all causes and

around 22% for post-neonatal mortality for respira-tory diseases Regarding damage in foetal health, although results are not always consistent, most studies show associations with exposure to air pol-lution during pregnancy However, the precise mechanisms of action of air pollutants on adverse reproductive results are still unknown, so is the per-iod of exposure most relevant during pregnancy and the specific pollutant which may represent a higher risk Follow-up studies evaluating personal exposure

to different air pollutants are required, allowing for the adequate evaluation of the impact of each pol-lutant in different periods of pregnancy, as well as providing hypotheses on their possible mechanisms

of action

Key words: Ambient air pollution, Congenital defects, Intrauterine growth retardation, Low birth weight, Mortality, Preterm delivery

Abbreviations: BW = birth weight; CI = confidence interval; CO = carbon monoxide; IQR = interquartile range; IUGR = intrauterine growth retardation; LBW = low birth weight; NOx = nitrogen oxides; O3= ozone; PAH = polycyclic aromatic hydrocarbons; PCB = polichlorinated biphenyls; PM10 = parti-cles£ 10 lm diameter; PM2.5= particles £ 2.5 lm diameter; TSP = total suspended particles; SGA = small for gestational age; SIDS = sudden infant death syndrome; SMR = standardized mortality ratio; SO2 = sulphur dioxide; WP = weeks of pregnancy

Introduction

Foetal growth may be altered by maternal

patholo-gies (diabetes, hypertension, etc.), by deficient diets,

by exposure to toxic substances (tobacco, alcohol,

drugs), and by ambient pollutants in air [1–3], in

water and in soil [4, 5]

The foetus and the infant present a special

vul-nerability, compared to adults, regarding ambient

toxicants due to differences in exposure, physiological

immaturity, and longer life expectancy after

expo-sure Results from epidemiological and experimental

studies show that foetuses and infants are especially

susceptible to the toxic effects of pollutants such as

suspended particles, polycyclic aromatic

hydrocar-bons (PAH), and tobacco smoke [6] In the case of

exposure to air pollutants where exposure occurs through inhalation, children inhale a relatively higher volume of air than adults

In recent years there is a growing concern about the possible influence on health of the exposure to air pollutants during pregnancy or first childhood; exposure to concentrations which may be reached nowadays in our homes or streets Recent studies have added proofs of the impact of exposure to air pollution on the risk of intrauterine or post-neonatal death [7, 8], or congenital defects [9], prematurity [10, 11] and foetal development [1, 3, 12]

We review the original studies which have evalu-ated the possible impact of ambient air pollution on the foetus and the first year of life, published in the last 10 years, from January 1994 to December 2003

Methods and materials

Search strategy

A bibliographical search was carried out in the online

database MEDLINE (http://www.ncbi.nlm.nih.gov/

entrez/query.fcgi) With the MESH Thesaurus, the

following search syntax was used (‘Air

Pollu-tion’[MeSH] AND (‘Pregnancy’ OR ‘reprod*’ OR

‘infant’ OR ‘foetal’) AND ‘Exposure’ Limits: 10

Years, Human) As a time frame, 10 years previous to

the search date (December 2003) were selected

Additional to the search in MEDLINE, a search in

the bibliographical database of the authors was

car-ried out and in the references of the selected articles,

The study by Bobak and Leon, 1992 [13] was

in-cluded for its relevance, despite having been

pub-lished before the period considered in this revision

Inclusion criteria The articles included follow these criteria: (a) original article; (b) observational epidemiological study; (c) exposure to outdoor air pollutants; (d) prenatal or up

to first year of life exposure, and (e) languages: English, French, Spanish, Portuguese or Italian Exclusion criteria

Articles which only dealt with (a) passive exposure to tobacco smoke; (b) exposure to indoor air pollutants,

or (c) working place exposure, were excluded Comparison of individual estimates

In order to facilitate comparison between studies, the odds ratios or risks ratios showed in Figures 1a and b

Study

3rd trimester

Retrospective Cohort

Annual

Population based Case-control

Annual Lin et al ,2001 Geographical

Exposure

Design

Annual

Geographical Landgren, 1996

0,5

1

1,5

2

2,5

3

3,5

4

4,5

area HC

OR

RR/OR

0,6

0,8

1

1,2

1,4

1,6

1,8

SO2 TSP

Exposure

Study

Design

Bobak & Leon 1999

Geographical Bobak, 2000 Maisonet et al, 2001 Annual

Case-control nested in a cohort

Whilhelm et al, 2003

Annual Wang et al, 1997

Prospective

cohort

3rd trimester

Maroziene &

Grazuleviciene, 2002

Annual

(a)

(b)

Figure 1 (a) Risk of low birth weight in studies evaluating its association with exposure to different ambient air pollutants (measured as an increase of 10lg/m3

of TSP, PM10, SO2or NOxand as an increase of 1 mg/m3de CO) during pregnancy (b) Risk of low birth weight in studies evaluating its association with exposure to ambient air pollution (exposed vs control area) during pregnancy

and 2 were recalculated to obtain the estimated effect

of each outcome for every increase in the levels of

TSP, PM10, SO2and NOxof 10lg/m3

, and of 1 mg/

m3in the levels of CO

Meta-analysis

In general, for most of the outcomes at study a very

scarce number of studies met with similar criteria for:

outcome, exposure or design So a formal

meta-analysis could not be attempted

However, for the case of infant mortality,

espe-cially for post-neonatal mortality, and for low

birth weight some studies included quite likely

mea-sures of exposure and, therefore, an approximation

to some overall estimates was done Most of these

studies included an indicator of particulates but not

always the same, so, in order to have comparable

measures, we approximate different levels of

expo-sure to 10lg/m3

of PM10 by using the following correction factors:

PM10 ¼ 0:6
TSP

PM2:5¼ 0:7
PM10:

Both corrections have been used by different agencies

or programmes in order to dispose of comparable

measures for particles (see Apheis Report 3 in www/

appheis.net, in press) Particularly, the first correction

factor is used in Central Europe, and it is very

appropriate here, as we are converting TSP measures

from Check Republic

The quantitative summary of study specific results

was obtained by means of meta-analysis techniques,

using STATA statistical package The combined

estimates were obtained by weighted regression,

in which the weights were the inverse of the local

variances, i.e.: using the ‘fixed effect model’ [14]

Heterogeneity was checked by a chi-square test under

the fixed effect hypothesis and, if heterogeneity was

detected, the ‘random effect model’ was applied For the purposes of this analysis, heterogeneity was assumed to be present if p < 0.20, however, in all meta-analyses carried out, the value of p for hetero-geneity test was always above 0.20

Results and discussion

Using the above strategy, a total of 31 articles was obtained The adverse reproductive effects evaluated

in the selected articles were: intrauterine mortality, child mortality within the first year of life, birth weight, premature delivery, intrauterine growth retardation, congenital defects, (Table 1)

Low birth weight, intrauterine growth retardation and premature delivery

Low birth weight (<2500 g) and premature delivery (<37 weeks of gestational age) are considered important predictors of foetal, neonatal and infant

Table 1 Adverse reproductive effects evaluated in the selected studies

(n = 31)

Intrauterine growth retardation (weight at birth

< percentile 10 for gestational age and sex)

3

In several articles more than one adverse reproductive effect

is evaluated, therefore the amount of articles which evaluate specific results do not sum the total

0,8 0,9 1 1,1 1,2 1,3 1,4 1,5 1,6

Maisonet M et al, 2001 Wilhelm M et al, 2003

Maisonet M et al, 2001 Wilhelm M et al, 2003

RR/OR

Study

Figure 2 Relative risk (and 95% CI) of low birth weight in studies evaluating its association with exposure to PM10and

SO2(measured as an increase of 10lg/m3

) and CO (measured as an increase of 1 mg/m3) during pregnancy

185

mortality, as well as of infant morbidity [15–18] and it

could even be a risk factor for adult morbidity [19]

Therefore, the ambient factors which may contribute

to reducing the weight at birth are a great concern for

public health [20]

Birth weight and prematurity are highly related,

since the weight at birth reflects two major

physio-logical processes: the foetal growth rate and the

extension or duration of gestation Therefore, low

birth weight may be due to either a short gestation

period or an intrauterine growth retardation (birth

weight < percentile 10 for gestational age and sex)

[21, 22] or to a combination of both causes Low birth

weight may be considered as a foetal growth measure

if the analyses are adjusted by gestational age or if

LBW studies are restricted to full-term births [23], in

this revision all the studies on the effect of air

pollu-tion on LBW, adjusted the statistical models by

ges-tational age or analyses were restricted to full-term

births except for the study carried out by Bobak M,

2000 [24]

We found 12 studies evaluating the association

between exposure to outdoor air pollutants and low

birth weight (LBW), 2 evaluate very low birth weight,

3 birth weight (BW), 10 prematurity and 3 evaluate

intrauterine growth retardation (IUGR) The most

studied pollutants have been: total suspended

parti-cles (TSP), SO2 and CO (Tables 2 and 3) In most

studies the evaluation to exposure was carried out

assigning the levels of air pollutants (year, trimester

or month means) reported by air pollution

monitor-ing stations regardmonitor-ing the proximity to the residence

of the mother at the time of delivery In other studies

exposure was assigned according to the distance of

the place of residence to an industrial area; and only

in one case an individual and direct evaluation of

exposure was carried out In this study, performed by

Perera et al [12] they evaluated exposure to PAH

during the third trimester of pregnancy through

personal air sampling in a sample of 263

Afro-American and Dominican non-smoking women of

between 18 and 35-year-old, living in New York and

who were registered in gynaecology and obstetrics

clinics during week 20 of gestation Exposure to PAH

among Afro-American women was significantly

associated with a lower weight at birth and smaller

head circumference after adjusting for potential

confounders

In all the studies reviewed, except in two [25, 26], a

higher risk of low birth weight was observed,

signif-icantly associated to air pollution levels However,

there is no consistency regarding to which pregnancy

trimester could be more relevant and the specific

pollutant which may represent a higher risk

Figures 1a and 1b summarise the results from studies

which have evaluated the association between

expo-sure to air pollutants and LBW Figure 1a represents

the results from the 6 studies which evaluated the

association between low birth weight (LBW) with

annual or third trimester of gestation exposure to specific pollutants (measured as an increase of 10lg/

m3 of TSP, PM10, SO2, NOx or NO2 and as an increase of 1 mg/m3 de CO during pregnancy) Figure 1b shows the results of four studies evaluating exposure in a dichotomous way (exposed vs control area), observing with odds ratios higher than null value and significant in all cases except in the study

by Landgren [26]

Results from meta-analysis in studies which ana-lyse the effect of exposure to PM10, SO2and CO on low birth weight show what we indicated previously

in a summarised way (Figure 2) Combined estimates show that a 10lg/m3

increase of PM10 or SO2 (an-nual or third gestational trimester mean) is associated with a 1.6% (CI 95%, 1.0–2.2%) and 1.5% (CI 95%, 0.7–2.4%) increase in the risk of low birth weight (<2500 g), respectively On the other hand, a 1 mg/

m3 increase of CO (annual or third gestational tri-mester mean) is associated with a 21% (CI 95%, 7.0– 36%) increase of LBW risk

Regarding premature delivery (<37 gestational weeks), all the studies, except the one by Landgren [26], observe association, although sometimes very small, with exposure to air pollutants From the 10 selected studies, only 5 studied the effect associated to specific pollutants (Table 3), therefore, due to the scarcity of studies evaluating this outcome and also the variability among those studies regarding the relevant exposure period considered, we do not present the combined effects estimates

Intrauterine growth retardation was studied

in three articles [3, 24, 27] In two of them [3, 27], association with exposure to particles was found (PM10 and PM2.5) and in the second study also with PAH Dejmek et al [3] studied this effect by exposure in each month of pregnancy, observing association with exposure in the first month of gestation The authors highlight the idea that, possibly, the effect which exposure to particles supposes for risk of IUGR depends highly

on the concentration of toxic compounds contained

in these particles, rather than on the level of particles alone [3] In fact the PAH, especially those which are big molecules, are better absorbed by the fine particles

The mechanisms of action of air pollutants on these adverse reproductive effects are not accurately known today The same could be said about the most relevant critical periods of exposure during preg-nancy due to the lack of consistency between the different studies regarding the magnitude of the effect according to the mother’s exposure along the differ-ent trimesters of pregnancy However, some mecha-nisms of action have been postulated:

(1) Infection of the mother of different aetiologies

is an important causal factor of premature delivery [28] being the association with genitourinary infection the most documented [29], therefore, it can be

3 ,

3 for

Control variables

Prospective cohort

3 in

Socioeconomic characteristics,

two pollutants

3 in

Sociodemographic characteristics, reproductive factors,

Retrospective cohort

California, 1989–1993

Socioeconomic, characteristics,

3 in

187

Authors, year, [reference]

3 ,

3 for

Control variables

2000 [3]

1994–1998 n(TD)

(35.6) Averages

characteristics, reproductive

2000 [54]

2000 [2]

case-control study

3 )

Socioeconomic characteristics,

3 )

3 ):

3 ):

2001 [56]

characteristics geographical

O3

Socioeconomic characteristics, reproductive

O3

2001a [57]

Socioeconomic characteristics,

2001 [1]

(1.06–1.62) Second

3 increase

3 ,

3 for

Control variables

Socioeconomic characteristics, reproductive

3 in

3 in

Prospective cohort

3 range

3 )

ethnicity Among

Case-control nested

O3

Socioeconomic characteristics,

ethnicity, reproductive

>0.05) Third

delivery, Kaohsiung

Taiwan, 1995–1997

Socioeconomic characteristics, season,

3 during

1.88–34.34) An

3 in

3 in

189

3 ,

3 for

Prospective cohort

3 ,

3 ,

3 increase

Prematurity OR

3 increase

Socioeconomic factors,

3 in

O3

O3

O3

Socioeconomic characteristics,

3 in

3 in

Socioeconomic characteristics,

3 in

Authors, year [reference]

3 ,

3 for

2002 [63]

Socioeconomic characteristics

2003 [61]

Socioeconomic characteristics,

2003 [62]

O2

O2

O3

Socioeconomic characteristics,

Prematurity OR

2003 [58]

Case-control nested

O3

Socioeconomic characteristics, race/ethnicity, reproductive factors,

1.01–1.15), Third

191

lighted that the possibility of respiratory infections

associated with the inhalation of air pollutants during

pregnancy could be a causal factor of premature

delivery [30]

(2) Viscosity increase of blood and of plasmatic

fibrogen due to inflammatory processes of peripheral

air ways associated with air pollution [31], this could

lead to an alteration of the umbilical and placental

blood flow, tranplacental glucose and total insulin,

which determine foetal growth [11, 24, 32–34]

(3) Diminution of the provision of oxygen in the

uterus, as a result of a diminution in the capability of

transporting oxygen due to an increase of

carboxy-hemoglobin as a consequence of acute or chronic

exposure to carbon monoxide [20]

(4) The foetal toxicity of exposure to polycyclic

aromatic hydrocarbons (PAH) has been associated

with effects on DNA or its transcription, through the

formation of PAH–DNA adducts [11, 24, 33], which

may result in the activation of the apoptosis [34], or

the binding to receptors of placental growth factors,

resulting in the decreased exchange of oxygen and

nutrients [3]

(5) High exposure near to the end of gestation

may cause disturbances of the

pituitary-adrenocor-tico-placental system [35], with possible

anti-estro-genic effects which may lead to a foetal toxicity [4]

(6) NO2is capable of oxidising tissue components

(e.g., proteins and lipids) and of eliminating the

anti-oxidising protective systems of the organism

Increased lipidic peroxidation in the maternal or

foetal compartment has been associated to

prematu-rity It has been observed in experimental studies that

NO2during pregnancy induces lipid peroxidation in

the placenta, high post-implantation embryonic

lethality, and disturbances of postnatal development

[36]

(7) The pathogenesis of IUGR is produced by an

abnormal reaction between the trophoblast and the

uterine tissues within the first weeks of gestation,

therefore, an alteration of growth may result from

a suboptimal placentation and maternal

hemody-namic maladaptation [37] which could be due to

exposure to air pollutants during the first month of

gestation [3]

Congenital defects

Only in two articles association between exposure to

air pollutants and the risk of congenital defects has

been evaluated [9, 38] In the study by Dummer et al

[38] mortality by congenital defects in areas close to

incinerators and crematoriums is studied; observing

an increase of risk in both areas (see Table 4)

In another study, Ritz et al [9], in the cohort

of neonates and foetuses delivered in southern

Cali-fornia in 1987–1993, observed an increase of risk of

cardiac anomalies with the exposure to CO and O

during the second month of gestation, period which coincides with the genesis of the heart Adjusted odds ratios at levels of CO‡ 2.60 mg/m3

was 2.84 (1.15– 6.99) for defects in the ventricular wall; and at levels

of O3‡ 57.2 lg/m3

was 2.51 (0.99–6.37) for valve and aorta artery defects This study may be considered as

an important referent due to the strong association found, as well as for its population-based character-istics, accounting for a large number of individuals under study, availability of wide coverage covariables

at individual level and examining vulnerable periods

of gestation

Intrauterine and infant mortality Foetus and infant mortality forms a group of serious effects which has been related to exposure to air pollution during pregnancy and early stages of life Among the 10 studies identified, 2 have evaluated intrauterine mortality (stillbirth after week 28) as outcome, 5 have studied the relationship of exposure

to air pollution with infant mortality (during the first year of life) and the remaining 3 have included dif-ferent outcomes from the other two groups

A summary of these studies is provided in Table 4 Different designs have been used but, given the low frequency of the outcomes being studied; the infor-mation has been obtained in all cases from public registers and statistics In seven studies a comparison using aggregated population data was carried out, four of them performed comparisons between the mortality rates among different geographical units and the other three are time series studies which evaluated the short term effect of the variations in the levels of air pollutants on the number of deaths at the earliest stages of life Of the three studies with an individual basis, two have used a retrospective cohort design [7, 39] and the other one, cases and controls [40] Most studies use air pollution data from public registers and surveillance and control systems In two

of them exposure is evaluated based on the distance

to exact sources such as coke works [41] or inciner-ators and crematoriums [38] Only in the study in Me´xico City air pollution measurements from a monitoring station operated by the investigators were used [42] Particles (in different forms such as TSP, PM10 or PM2.5), SO2and nitrogen oxides have been included in seven of the eight studies which present data of pollutants CO and ozone have been included in two of the time series studies In the study in Sweden [26] the possible effect of hydro-carbons was examined and in the Sao Paulo study [8]

an index of overall pollution was developed to avoid certain problems from correlation between pollutants

in the statistical analysis

A common problem of these studies is that related with errors in the measurement of exposure, mainly due to differences between measurements from sta-tions and real exposure of each person in a given