ARTICLE IN PRESS Environmental Research 95 (2004) 106–115 Review A review of the literature on the effects of ambient air pollution on fetal growth Mildred Maisonet,a,* Adolfo Correa,b,c Dawn Misra,d,e and Jouni J.K. Jaakkolaf a Pan American Health Organization, Washington, DC, USA Department of Epidemiology, The Bloomberg School of Public Health, Johns Hopkins University, Baltimore, MD, USA c National Center on Birth Defects and Developmental Disabilities, Centers for Disease Control and Prevention, Atlanta, GA, USA d Department of Population and Family Health Sciences, The Bloomberg School of Public Health, Johns Hopkins University, Baltimore, MD, USA e Department of Health Behavior and Health Education, University of Michigan School of Public Health, Ann Arbor, MI, USA f Institute of Occupational Health, The University of Birmingham, UK b Received 3 February 2003; received in revised form 18 December 2003; accepted 5 January 2004 Abstract A systematic review of the literature on the effects of air pollution on low birth weight (LBW) and its determinants, preterm delivery (PTD) and intrauterine growth restriction (IUGR), was conducted. Twelve epidemiologic investigations that addressed the impact of air pollution on four pregnancy outcomes were identified. Results were analyzed separately for each perinatal outcome because of differences in pathogenic mechanisms. Effects of air pollution were apparent on PTD and IUGR, but not on LBW. Most of the associations reported were rather small. The estimation of summary effects was not meaningful because of the heterogeneity of the effect estimates arising from differences in the measurements of outcome, exposure, and confounders and the small number of studies per outcome (four studies for PTD and six for IUGR). Current scientific knowledge on the impact of air pollution on fetal growth is still limited; thus, several issues should be examined further. r 2004 Published by Elsevier Inc. Keywords: Literature review; Ambient air pollution; Fetal growth; Low birth weight; IUGR; Preterm delivery Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 2. Literature search . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 3. Characteristics of the studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 4. Results . . . . . . . . . . . . . . . 4.1. Low birth weight . . . . . . . 4.2. Very low birth weight . . . . 4.3. Preterm delivery . . . . . . . 4.4. Intrauterine growth restriction . . . . . 108 108 108 108 110 5. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1. Preterm delivery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2. IUGR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 111 112 6. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . *Corresponding author. 1017 Providencia Avenue, Box 9459, Santiago, Chile, 6643091. Fax: +562-264-9311. E-mail address: maisonetm@chi.ops-oms.org (M. Maisonet). 0013-9351/$ - see front matter r 2004 Published by Elsevier Inc. doi:10.1016/j.envres.2004.01.001 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ARTICLE IN PRESS M. Maisonet et al. / Environmental Research 95 (2004) 106–115 107 1. Introduction 2. Literature search Recent epidemiological studies have shown associations of low birth weight (LBW) and its determinants, preterm delivery (PTD) and intrauterine growth restriction (IUGR), with ambient air pollution (Bobak, 2000; Dejmek et al., 1999; Ha et al., 2001; Maisonet et al., 2001; Ritz and Yu, 1999; Ritz et al., 2000; Wang et al., 1997; Xu et al., 1995). Interpretations of these findings are difficult, in part, because LBW (birth weight o2500 g) represents a heterogeneous group of outcomes with different pathogenic mechanisms. Some infants have LBW as a result of PTD (o37 completed weeks of gestation at delivery), while others are a result of IUGR (birth weight less than that expected for a given gestational age). Some LBW infants are both preterm and growth restricted for their gestational age. Such subgroups of LBW may not necessarily arise from the same set of risk factors. Some maternal prenatal determinants may be associated with an increased risk of LBW through effects on the length of gestation alone (e.g., premature rupture of the membranes, placenta abruptio), others through effects on intrauterine fetal growth alone (e.g., maternal weight gain, hypertension), and some possibly through effects on both PTD and IUGR (e.g., maternal cigarette smoking, malformations). Whether the reported association reflects an effect of air pollution on PTD, IUGR, or both should be better understood. Ambient air pollutants also vary in nature (e.g., CO, SO2, and particulate matter) and possible health effects. Exposure to ambient levels of CO may result in the formation of carboxyhemoglobin in the mother, which in turn could result in decreased oxygen delivery to tissues, including the fetus. Inhaled particles, in contrast, have been reported to increase blood viscosity, which may have an adverse effect on placental function, thus restricting fetal growth (Ha et al., 2001). In the absence of definite biological mechanisms for the possible associations between some ambient air pollutants and fetal growth, it is important to determine whether the associations reported are consistent and whether they reflect the effect of a specific pollutant or a mixture of pollutants. However, thus far there have been no evaluations of whether such associations are for specific air pollutants, the consistency of the reported associations, and the levels of ambient air pollution at which such associations are evident. In this report, we review published studies on the association of ambient air pollutants with LBW, PTD, and IUGR. Because these outcome measures may have different pathogenic and etiologic mechanisms, we conducted this review by the type of outcome. The review was limited to reports published in the English language peer-reviewed scientific literature for four outcome measures: LBW, very low birth weight (VLBW), PTD (i.e., birth at o37 weeks of gestation), and IUGR. We selected reports that evaluated any of the following ambient air pollutants: CO, SO2, nitrogen oxides (NOx), particulate matter, and ozone. We identified articles through Medline searches, bibliographies of individual articles, and reviews of scientific journals from 1966 through December 2001. 3. Characteristics of the studies Twelve epidemiologic investigations that addressed the impact of air pollution on four pregnancy outcomes were identified, with some studies examining multiple outcomes (Table 1). The outcomes assessed were LBW (4 studies), VLBW (2 studies), PTD (4 studies), and IUGR (6 studies). Many of the studies shared similar design and methodologic features. The 12 investigations consisted of 9 population-based cross-sectional studies, 1 case-control study, 1 case-cohort study, and 1 ecologic study. Most often data on outcome and confounders were ascertained from vital records. Some studies used selfadministered maternal questionnaires and medical records to collect information, and one ascertained information by means of personal interviews. All studies ascertained and adjusted for confounding in the analyses. The extent to which confounding was addressed varied among studies and was influenced by the source of the data. Among the studies that were based on data from vital records, the assessment of confounding consisted more often on adjustments for maternal, and infant characteristics. Studies using other data sources were able to address other potential confounders, such as smoking, alcohol use, and weight gain during pregnancy. In all but one of the studies the exposure assessment was based on data from stationary air pollution monitors that provided estimates of the concentrations of pollutants in geographically defined regions. The exception was the sole case-control study, which instead employed environmental transport modeling to estimate exposures for the home where the mother resided at the time of the birth of her infant. Particulate matter and SO2 were the ambient contaminants most commonly evaluated. Nine studies examined some measure of particles, including total suspended particles (TSP), respirable particles (PM10), and ultrafine particles (PM2.5). Exposure to SO2 was examined in eight studies. Five studies evaluated ambient NOx levels, five evaluated ambient CO levels, and two evaluated ozone. ARTICLE IN PRESS M. Maisonet et al. / Environmental Research 95 (2004) 106–115 108 Table 1 Characteristics of the studies by study design First author (year) Country Case-cohort study Alderman et al. (1987) Case-control study Rogers (2000) Outcome Size of study group Exposure variable United States LBW 988 LBW 1872 normal weight Average concentrations of CO during third trimester United States VLBW 143 cases 202 controls Combined average annual concentration of TSP and SO2 during year of birth 25,370 38,718 Average concentrations of TSP and SO2 7 days before birth Above and below the mean SO2 and NO values of all municipalities; the municipality with the highest exposure level was compared to other municipalities Average concentrations of TSP and SO2 during third trimester Average concentrations of PM2.5 and PM10 on each of 9 consecutive 30-day periods after effective date of conception Average concentration of CO during third trimester Population-based cross-sectional studies Xu (1995) China Preterm birth Landgren (1996) Sweden Preterm birth, very preterm birth, LBW and VLBW Wang (1997) China Term LBW 74,671 Dejmek (1999) Czech Republic SGA 1943 Ritz and Yu (1999) Bobak (2000) United States Term LBW 125,573 Czech Republic 108,173 Ritz (2000) United States LBW, preterm birth, and SGA Preterm birth 143,196 Ha (2001) Korea Term LBW 286,474 Maisonet (2001) United States Term LBW 89,557 Ecologic study Bobak and Leon (1999) Czech Republic LBW 45 residential districts Average concentrations of TSP, SO2, and NOx during first, second, and third trimesters Average concentrations of PM10, CO, NO, and O3 1, 2, 4, 6, 8, 12, and 26 weeks before birth and during first and second month of pregnancy Average concentrations of TSP, SO2 NO2, CO, and O3 during first and third trimesters Average concentrations of PM10, SO2,and CO during first, second, and third trimesters Average annual concentrations of TSP, SO2, and NOx during year of birth LBW, low birth weight; VLBV, very low birth weight; SGA, small-for-gestational-age; TSP, total suspended particles; PM, particulate matter. 4. Results 4.1. Low birth weight Four studies evaluated LBW as an outcome (Table 2). The contaminants assessed were TSP, SO2, CO, and NOx. One of three studies that assessed the effects of exposures to SO2 reported an increase in risk of LBW (Odds ratio (OR), 1.10 per 50 mg/m3 increase; 95% confidence interval (CI), 1.01–1.20) (Bobak and Leon, 1999). Exposures to TSP, CO, or NOx did not appear to be associated with increases in LBW. 4.2. Very low birth weight Two studies evaluated VLBW as an outcome (birth weight o1500 g) (Table 2). The contaminants assessed were TSP, SO2, and NOx. In one study exposure was measured by combining both TSP and SO2 data (Rogers et al., 2000). An increased risk of VLBW (OR, 2.88; CI, 1.16–7.13) for the group at the highest level of exposure (456.75 versus o9.94 mg/m3) was reported for these combined measure. The second study looked at the independent effects of SO2 and NOx (Landgren, 1996). No increase in risk was reported for these pollutants. 4.3. Preterm delivery Four studies examined PTD as an outcome (Table 3). The contaminants assessed in these studies were particulate matter, SO2, CO, and NOx. The three studies that assessed the effect of exposure to particulate matter, measured either as TSP or as PM10, reported positive results for this association. Two of them reported that exposures, late in pregnancy, to particulate matter increased the risk for PTD. Odds ratios of 1.10 (per 100-mg/m3 increase in TSP; CI, 1.01–1.20) (Xu et al., 1995) and 1.19 (per 50-mg/m3 increase in PM10; CI, 1.01– 1.40) (Ritz et al., 2000) were reported. The third study reported increases in the risk of PTD only for maternal exposures occurring during the first trimester of pregnancy (OR, 1.18 per-50 mg/m3 increase in TSP; CI, 1.05–1.31) (Bobak, 2000). Two of three studies that assessed the effects of maternal exposure to SO2 on PTD reported positive Table 2 Results of studies of ambient air pollutants and low birth weight (LBW) and very low birth weight (VLBW) First author (year) Outcome and exposure measures Exposure categories Odds ratio (95% CI) Covariates/stratified variables Alderman et al. (1987) LBW and average concentration of CO in the third trimester X3 ppm vs. o1 ppm 1.5 (0.7–3.5) Maternal race, education Landgren (1996) LBW and mean concentrations of SO2 and NOx for municipalities Rogers (2000) LBW and average concentrations of TSP, SO2, 50-mg/m3 increments in TSP and NOx in first, second, and third trimesters First trimester Second trimester Third trimester 50-mg/m3 increments in SO2 First trimester Second trimester Third trimester 50-mg/m3 increments in NOx First trimester Second trimester Third trimester VLBW and combined average annual concentration of TSP and SO2 at the birth home during the year of birth 0.97 (0.73–1.29) 0.99 (0.88–1.11) 0.95 (0.75–1.22) 0.98 (0.77–1.25) 1.46 (0.80–2.65) 0.97 (0.73–1.29) 50-mg/m3 increments TSP SO2 NOx Bobak (2000) 0.92 (0.83–1.01) 0.95 (0.86–1.05) Mean income, mean savings, mean number of people/car, proportion of births outside of marriage, proportion of divorces to new marriages, legally induced abortions/100 live births, proportion of gypsies in the population 1.03 (0.95–1.11) 1.10 (1.01–1.20) 0.99 (0.89–1.10) Month of birth, gestational age, gender, parity, maternal age, education, marital status, nationality 1.13 (0.93–1.38) 1.14 (0.92–1.40) 1.14 (0.93–1.38) 1.01 (0.88–1.17) 0.95 (0.82–1.10) 0.97 (0.85–1.10) 0.98 (0.81–1.18) 0.99 (0.80–1.23) 0.97 (0.80–1.18) TSP+ SO2 0.99 (0.51–1.72) 1.27 (0.68–2.37) 2.88 (1.16–7.13) 109 9.9–25.1 mg/m3 25.2–56.8 mg/m3 456.8 mg/m3 Maternal race, age, weight gain, prepregnancy weight, toxemia, alcohol and drug use, cigarette smoke, stress, prenatal care, income, parents’ education, gender ARTICLE IN PRESS LBW and average annual concentrations of TSP, SO2, and NOx during the year of birth Year of birth, parity, maternal age M. Maisonet et al. / Environmental Research 95 (2004) 106–115 Bobak and Leon (1999) Above or below the contaminant’s mean exposure value for all of the municipalities SO2 NOx Municipality with the highest pollutant levels vs. all others SO2 NOx VLBW (o1500 g ) and mean concentrations of Above or below the contaminant’s mean SO2 and NOx for municipalities exposure value for all of the municipalities SO2 NOx Municipality with the highest pollutant levels vs. all others SO2 NOx ARTICLE IN PRESS 110 M. Maisonet et al. / Environmental Research 95 (2004) 106–115 Table 3 Results of studies of ambient air pollutants and preterm delivery (PTD) First author (year) Outcome and exposure measures Xu (1995) PTD and average concentrations of 100-mg/m3 increments in TSP TSP and SO2 during the 7 days Each ln-mg/m3 increment in SO2 before delivery Landgren (1996) PTD and mean concentrations of SO2 and NOx for municipalities Exposure categories Above or below the contaminant’s mean exposure value for all of the municipalities SO2 NOx Municipality with the highest pollutant levels vs. all others SO2 NOx Very PTD (o32 weeks) and mean Above or below the contaminant’s concentrations of SO2 and NOx for mean exposure value for all of the municipalities municipalities SO2 NOx Municipality with the highest pollutant levels vs. all others SO2 NOx Bobak (2000) Ritz (2000) PTD and average concentrations of First trimester 50-mg/m3 increments TSP, SO2, and NOx in first, second, and third trimesters 50-mg/m3 increments 50-mg/m3 increments Second trimester 50-mg/m3 increments 50-mg/m3 increments 50-mg/m3 increments Third trimester 50-mg/m3 increments 50-mg/m3 increments 50-mg/m3 increments Covariates/stratified variables 1.10 (1.01–1.20) 1.21 (1.01–1.46) Gender, maternal age, residential area, temperature, humidity, day of week, season Year of birth, parity, maternal age 0.97 (0.88–1.07) 1.00 (0.90–1.10) 0.90 (0.66–1.22) 1.09 (0.97–1.21) 0.95 (0.75–1.21) 0.89 (0.70–1.13) 1.16 (0.60–2.26) 0.96 (0.72–1.27) in TSP in SO2 in NOx 1.18 (1.05–1.31) 1.27 (1.16–1.39) 1.10 (1.00–1.21) in TSP in SO2 in NOx 1.11 (0.97–1.26) 1.25 (1.14–1.38) 1.08 (0.98–1.19) in TSP in SO2 in NOx 1.12 (0.97–1.28) 1.24 (1.13–1.36) 1.11 (1.00–1.23) PTD and average concentrations of First month of pregnancy 50-mg/m3 increments in PM10 PM10, CO, NO, and ozone during the first and second months of 3-ppm increments in CO pregnancy and 1, 2, 4, 6, 8, 12, Six weeks before birth and 26 weeks before birth 50-mg/m3 increments in PM10 3-ppm increments in CO findings. One of the studies reported similar effects for each trimester’s exposure measure with the OR ranging between 1.27 (per 50-mg/m3 increase; CI, 1.16–1.39) in the first trimester and 1.24 (CI, 1.13–1.36) in the third trimester (Bobak, 2000). The other study reported an increase in the risk of PTD with exposures to SO2 during the last week of pregnancy (OR, 1.21 per ln-mg/m3 increase; CI, 1.01–1.46) (Xu et al., 1995). Two studies addressed the impact of exposures to NOx on the risk of PTD. One reported an increase in the risk of PTD with exposures occurring during the first (OR, 1.10 per 50-mg/m3 increase; CI, 1.00–1.21) and third trimester (OR, 1.11; CI, 1.00–1.23) (Bobak, 2000). Odds ratio (95% CI) 1.12 (0.97–1.29) 1.03 (0.96–1.10) Month of birth, gender, parity, maternal age, education, marital status, nationality Gender, parity, maternal age, race, education, prenatal care, interval since previous birth, tobacco smoking, previous LBW or preterm 1.19 (1.01–1.40) 1.05 (0.97–1.10) Only one study assessed the effect of maternal exposure to CO on PTD. No effects were reported for this exposure. 4.4. Intrauterine growth restriction Six studies examined IUGR. The contaminants assessed in these studies were particulate matter, SO2, CO, NOx, and O3. Three of five studies reported positive results for exposures to particulate matter. One reported effects with exposures in the third trimester of gestation (OR, 1.10 per 100-mg/m3 increase in TSP; CI, 1.05–1.14) (Wang et al., 1997), while two others reported effects for ARTICLE IN PRESS M. Maisonet et al. / Environmental Research 95 (2004) 106–115 exposures occurring in the early stages of pregnancy. A 2.64-fold (CI, 1.48–4.71) (Dejmek et al., 1999) increase in the risk of IUGR was reported in association with exposures to particles during the first month of pregnancy among births from the highest PM10 exposure category (450 versus o40 mg/m3). A smaller effect on risk was reported for exposures during the first trimester (OR, 1.04 per trimester interquartile increase in TSP; CI, 1.00–1.08) (Ha et al., 2001). Three of four studies reported positive results for exposures to SO2 at different stages of pregnancy. One study reported effects with exposures occurring in the first trimester (OR, 1.06 per interquartile increase; CI, 1.02–1.10) (Ha et al., 2001). Another reported increases in risk ranging from 1.13 (CI, 1.05–1.22) to 1.18 (CI, 1.12–1.25) for second-trimester exposures falling within the 25th and 95th percentile of the exposure distribution (Maisonet et al., 2001). A third study reported an increment in risk with exposures occurring in the third trimester (OR, 1.11 per 100–mg/m3 increase; CI, 1.06– 1.16) (Wang et al., 1997). All three studies that assessed the relationship between IUGR and CO reported positive results. Two reported effects for exposures occurring in the third trimester of pregnancy with odds ratios of 1.22 (CI, 1.03–1.44) for births in the highest exposure category (45.5 versus o2 ppm) (Ritz, 2000) in one study and 1.31 (per 1-ppm increase; CI, 1.06–1.52) (Maisonet et al., 2001) in the other. The third study reported an increase in risk with exposures in the first trimester (OR, 1.08 per interquartile increase; CI, 1.04–1.12) (Ha et al., 2001). One of two studies that examined the association between NOx exposures and IUGR reported positive findings, specifically with exposures occurring during the third trimester (OR, 1.07 per interquartile increase; CI, 1.03–1.11) (Ha et al., 2001). This same study was the only one to assess the effects of O3, and it reported positive results for exposures occurring during the third trimester (OR, 1.09 per interquartile increase; CI, 1.04– 1.14) (Ha et al., 2001). 5. Discussion The studies reviewed for this analysis reported positive statistical associations between different air pollutants and LBW, VLBW, PTD, and IUGR. Each outcome was assessed separately because of differences in pathogenic mechanisms. Overall, the number of studies that were available in the literature suggests that the scientific knowledge of the impact of air pollution on LBW and its determinants is still limited. These were but a few studies available for each outcome, which limited our potential to make inferences about any individual outcome: IUGR, six studies; PTD, four studies; LBW, four studies; VLBW, 111 two studies. Such was the case for VLBW, for which only two studies had been conducted. Also, assessments focused on effects at the tails of the birth-weight distribution curve when examining LBW and VLBW. The change in birth weight, analyzed as a continuous outcome variable, with exposure to air pollution was briefly addressed in two studies. For this reason, the meaning of the associations reported for these two outcomes is not further addressed in our discussion. Effects of air pollution were apparent on PTD and IUGR, but not for LBW. As mentioned earlier, LBW represents a heterogeneous group of outcomes, and this finding may reflect the lack of true associations with such a heterogeneous outcome. Contaminants assessed and the exposure parameters constructed for them varied across studies. For instance, contaminants assessed in one study were not necessarily assessed in all others. The exposure assessment included the construction of both continuous and categorical measures, and continuous measures were estimated for different increases in exposure levels across studies. Finally, the periods of exposure for which effects were assessed also varied across studies. Because of the heterogeneity of the exposure measure, as well as that of the outcome and confounders measures, the calculation of summary effects estimates was not meaningful. The small number of studies per outcome limited the possibility for conducting subgroup analysis. Individuals are most likely exposed to more than one contaminant, since a single source (e.g., car exhaust) may produce more than one contaminant. Because positive results from individual effects may be a surrogate of more complex exposures and it is not feasible to make an evaluation of independent effects by outcome, our discussion refers to the impact of air pollution as a whole. 5.1. Preterm delivery There was little variation in the magnitude of the positive associations reported for PTD, and estimated effects were rather small (ORs, 1.10–1.27). All studies treated PTD as a dichotomous variable defined as a birth at less than 37 weeks of gestation. Gestational age is commonly estimated by counting the number of weeks in between the first day of the last menstrual period (LMP) and the date of birth. One study reported using this method to estimate gestational age (Bobak, 2000). Another study reported having information on the LMP date, but did not specifically state that the estimation of gestational age was based on the LMP date (Xu et al., 1995). The other studies did not specify how gestational age was calculated (Landgren, 1996; Ritz et al., 2000). Since outcome data for the latter were obtained from vital records, it is likely that the same method was employed. ARTICLE IN PRESS 112 M. Maisonet et al. / Environmental Research 95 (2004) 106–115 The estimation of gestational age based on the LMP date is as good as those obtained through other methods, such as ultrasonography. Errors in the reporting of the LMP date, though, occur when women are uncertain of the date of the LMP or are mistaken about it. Poor recall, postconception bleeding, and menstrual irregularities are some reasons for an erroneous reporting (Kline, 1989). Systematic misclassification would be a concern if the distribution of these and other situations that influence the reporting of the LMP date was similar to that of the exposure. Assuming that all studies used the LMP date to estimate gestational age, it is unlikely that systematic misclassification would explain the results of the individual studies or their differences. Any determinant of PTD becomes a confounder when its distribution in the study group is related to the studied exposure. The studies reporting positive results used logistic regression techniques to address confounding (Xu et al., 1995; Bobak, 2000; Ritz et al., 2000). The choice of PTD determinants assessed for potential confounding effects varied across studies. Differences in sources of data across studies influenced the availability of data for the assessment of confounding. Whether uncontrolled factors are independent of the exposure distribution may prove difficult to establish and vary by pollutant. It may be that there is little correlation between determinant’s of the study outcomes and air pollution for contaminants the spatial distributions of which could be homogeneous within an area, such as PM10 and SO2. In contrast, pollutants like CO and NOx are not as homogeneously distributed. Since automobile emissions are the main source of these contaminants, their spatial variations are determined mainly by traffic density. In such situations confounding could occur through the determinant’s correlation to factors that increase the possibility of living in areas of high traffic density, such as socioeconomic conditions. Exposure misclassification may explain the null effects reported in one study of PTD. Landgren (1996) constructed a qualitative exposure measure, categorizing births as above or below a mean value, while the three other studies used the mothers’ average exposure concentrations during pregnancy. The latter is a better surrogate of absorbed dose. A better understanding of the induction period for PTD could help elucidate the biologic plausibility of the associations reported. Earlier research on potential etiologic factors for PTD has emphasized exposures in late pregnancy (late second to early third trimesters) such as infections and stress (Lockwood, 1994). One might hypothesize that the induction period is short and immediate in nature, leading to the occurrence of delivery soon after a relevant exposure has occurred. However, long-term exposures could potentially build up, reaching a threshold and triggering labor. Theore- tically, these two types of effects could take place concurrently with different levels or combinations of exposure. The results suggest the potential for both types of effects. One study used a sliding-scale approach to assess effects on different exposure periods throughout pregnancy (Xu, 1995). Because regression coefficients reached the maximum values when the exposure was lagged 7–8 days, the period during the 7 days previous to delivery was considered the critical time during which an exposure could cause the effect. Another study averaged pollutant measurements for the periods of 1, 2, 4, 6, 8, 12, and 26 weeks before birth and the first and second months of pregnancy (Ritz et al., 2000). Positive results were reported only for exposures occurring during the period of 6 weeks before birth. In contrast to these two studies, which reported significant effects of late-pregnancy exposures, a third study detected effects of early pregnancy exposures only (Bobak, 2000). In this study exposures were assessed for the entire pregnancy by developing average exposure measures for each trimester. Studies of maternal smoking have reported positive associations between this habit and PTD and premature rupture of the membranes (McDonald et al., 1992; Williams et al., 1992). Smoking also increases the risk of placental abruption, which is a cause of induced and spontaneous PTD (Ananth et al., 1996). Some components of air pollution may have effects similar to those of cigarette smoking. 5.2. IUGR Five of six studies reported positive effects of ambient air pollution on IUGR. Four studies reported small effects, with odds ratios ranging from 1.04 to 1.22, while one study reported moderate effects (OR, 2.6) (Table 4). An important element in the assessment of the impact of air pollution on IUGR is the way in which this outcome was measured in the studies. Four studies compared infants born at a LBW at term (437 weeks of gestation) with other-term infants (Wang et al., 1997; Ritz and Yu, 1999; Ha et al., 2001; Maisonet et al., 2001). This assumed that 37 weeks of gestation was sufficient to reach 2500 g (5.5 pounds) and that a failure to reach this weight was indicative of IUGR. Another dichotomous measure of IUGR was small-for-gestational age (SGA), for which the proportion of infants under the 10th percentile of birth weight for a particular gestational age stratified by gender was identified (Dejmek et al., 1999; Bobak, 2000). All studies of term LBW used the same definition: a live birth weighting less than 2500 g and a gestational age above 37 weeks. Birth weight can be objectively measured, and one would expect that its measurement would be consistent across studies. All four studies used ARTICLE IN PRESS M. Maisonet et al. / Environmental Research 95 (2004) 106–115 113 Table 4 Results of studies of ambient air pollutants and intrauterine growth restriction (IUGR) First author (year) Outcome and exposure measures Exposure categories Odds ratio (95% CI) Wang (1997) Term LBW and average concentrations of TSP and SO2 in the third trimester 100-mg/m3 increments in TSP 100-mg/m3 increments in SO2 1.10 (1.05–1.14) Gestational age, year of birth, gender, 1.11 (1.06–1.16) maternal age, residential area Dejmek (1999) Small for gestational age and average PM10 in first month of gestation 40–49 vs. o40 mg/m3 concentrations of PM10 and PM2.5 in each month of gestation X50 vs. o40 mg/m3 PM2.5 in first month of gestation 27–36 vs. o27 mg/m3 X37 vs. o27 mg/m3 Year, maternal education, marital 1.62 (1.07–2.46) status, smoking habits, maternal 2.64 (1.48–4.71) height, prepregnancy weight, season Ritz and Yu (1999) Term LBW and average concentration of CO in the third trimester 2.2–5.4 ppm vs. o2 ppm X5.5-ppm vs. o 2 ppm 1.04 (0.96–1.13) Gestational age, gender, maternal age, 1.22 (1.03–1.44) race, education, prenatal care, interval since previous birth, travel time to work, walking to work Bobak (2000) Small for gestational age and average concentrations of TSP, SO2, and NOx in first, second, and third trimesters 50-mg/m3 increments in TSP in first trimester 50-mg/m3 increments in SO2 in first trimester Gestational age, month of birth, 0.89 (0.75–1.06) gender, parity, maternal age, education, marital status, nationality 0.91 (0.80–1.04) Ha (2001) Term LBW and average concentrations of PM10, SO2, NO2, CO2, and O3 in first and third trimesters First trimester interquartile increase TSP SO2 NO2 CO Third-trimester interquartile increase O3 Maisonet (2001) 1-ppm increment in CO Term LBW and average concentrations of PM10, SO2 and CO First trimester in first, second, and third trimesters Second trimester Third trimester SO2 in second trimester 6.5–8.8 8.9–11.9 12.0–18.2 the LMP date to estimate gestational age. As we mentioned earlier, this estimation is prone to errors; however, it is unlikely that it varies systematically across studies. Both studies that estimated SGA were from the Czech Republic. One study defined SGA as a birth weight that fell below the 10th percentile by gender and gestational week for live births in the Czech Republic between 1991 and 1993 (Dejmek et al., 1999). Gestational age was estimated by gynecologists using the LMP date and other available information, such as ultrasound data. There was no mention of whether an algorithm was developed for this calculation. The other study used the same definition of SGA but did not indicate the fetal growth standard (Bobak, 2000). Gestational age was estimated from the LMP date alone. Both studies were from the Czech Republic, but one did not give enough Covariates/stratified variables 1.26 (0.81–1.95) 2.11 (1.20–3.70) Gestional age, gender, parity, maternal age, education 1.04 (1.00–1.08) 1.06 (1.02–1.10) 1.07 (1.03–1.11) 1.08 (1.04–1.12) 1.09 (1.04–1.14) Gestional age, gender, parity, maternal 1.08 (0.91, 1.28) age, race, education, marital status, 1.14 (0.83, 1.58) prenatal care, smoking and alcohol 1.31 (1.06, 1.52) use, weight gain, previous terminations, season 1.18 (1.12, 1.25) 1.12 (1.07, 1.17) 1.13 (1.05, 1.22) information to determine whether they employed the same growth charts. An assessment of whether one chart or the other would lead to more births being classified as IUGR was not possible; however, any variations in classification were likely to be random, and the impact would be to bias the effect estimates toward the null value of 1. In the absence of a standard population, term LBW is estimated as an alternative way to separate the effect of PTD in the assessment of the effects of air pollution on fetal growth. While both attempt to measure fetal growth, term LBW and SGA are not necessarily the same outcomes. Term LBW focuses on term infants and SGA uses the smallest birth at all gestational ages. The use of term LBW as the study outcome may present some problems for evaluating the effects of air pollution on fetal growth. Positive effects were reported ARTICLE IN PRESS 114 M. Maisonet et al. / Environmental Research 95 (2004) 106–115 by four studies assessing this outcome; however, even with the exclusion of PTD, term LBW at term gestation still represents a heterogeneous outcome. Term LBW infants may include, for example, incorrectly classified preterm births. In addition, the exclusion of infants with PTD may impact the study findings if these births are not independent of the exposure. In contrast to the studies on term LBW, which reported small effects, one study on SGA reported moderate effects (OR, 2.6) (Dejmek et al., 1999). One study is not sufficient to reach conclusions about the association between air pollution and SGA; however, the increase in the magnitude of the association may reflect an effect of air pollution on growth restriction rather than PTD. Adjustments for confounding in assessing the effects of air pollution on IUGR varied across studies. Some studies had data on maternal and infant characteristics (Bobak, 2000; Ha et al., 2001); others had information on and adjusted for other determinants (Dejmek et al., 1999; Maisonet et al., 2001). For reasons mentioned earlier, it is not clear which variables are likely to be real confounders and it is difficult to predict how results differ with adjustments for confounders or the lack thereof. At 28 weeks, the fetus has gained only about 32% of its average term birth weight, gaining the rest of its birth weight in subsequent weeks (Kline, 1989). Since most of the increase in weight occurs during the third trimester, one would hypothesize that this is the period of time when a relevant exposure is more likely to have an adverse effect on birth weight. Studies on maternal smoking are consistent with this hypothesis: a study focusing on the timing of exposure found little effect for smoking on birth weight if women quit prior to the third trimester but substantial decreases in birth weight if women continued to smoke (Lieberman et al., 1994). The effects of air pollution on IUGR were evaluated at various time periods during pregnancy across the studies. Three studies reported results that were consistent with the hypothesis that third-trimester exposures may have an adverse effect on IUGR (Maisonet et al., 2001; Ritz and Yu, 1999; Wang et al., 1997). Two other studies reported effects from exposures early in pregnancy. One study found an increased risk of IUGR with exposures in the first month of pregnancy (Dejmek et al., 1999), while another reported effects with first-trimester exposures. The reporting of effects with exposures at the early stages of pregnancy suggest that the mechanism through which air pollution has an adverse effect on birth weight is not the same as that for cigarette smoking. There may be more than one mechanism of effect acting concurrently or a more chronic-type effect. Since this is not clear, it would be useful to conduct studies that differentiate the outcomes seen for each scenario: asymmetric SGA (i.e., low weight for height), suggesting late effects, and symmetric SGA (i.e., normal weight for height), suggesting more chronic effects. Transplacental exposure to polycyclic aromatic hydrocarbons (PAHs) from maternal inhalation of PAHs adsorbed to air particles may explain an adverse effect of particulate matter on fetal growth. Perera and colleagues (1998) reported that infants with higher PAH–DNA adduct levels from umbilical cord leukocytes had decreased birth weight, length, and head circumference compared to infants with lower PAH– DNA adducts. The mechanisms by which PAHmediated exposures exert an effect on fetal growth have not been identified; disruptions to both the endocrine and the nervous systems are some possibilities. Another potential mechanism involves the hematologic effects of air pollution. Inhaled particles are able to provoke alveolar inflammation, with the release of mediators capable of increasing blood coagulability, leading to an increase in blood viscosity during episodes of air pollution (Seaton et al., 1995). It has been suggested that changes in viscosity can influence blood perfusion and have an adverse effect on placental functions, thereby restricting fetal growth (Bobak, 2000; Ha et al., 2001). CO is another ambient pollutant that may mediate the effect of air pollution on fetal growth. CO competes with oxygen (O2) for hemoglobin binding sites, blocking its O2-carrying capacity and reducing the amount reaching body tissues. The potential effects of air pollutants on many systems suggest that the fetus becomes vulnerable to adverse effects of air pollution at conception. 6. Summary Increases in ambient air pollution may increase the risk of PTD and IUGR among live births. Overall, the effect of air pollution on these outcomes is smaller than the effects of other known risk factors. However, it is not uncommon to find levels of ambient air pollutants in urban areas that are similar to those observed in the studies that reported positive results. This would be a cause for concern if the prevalence of exposure to these levels is elevated, since the public health impact would be substantial even if the causal effect were small. The estimation of summary effects was not meaningful because of the heterogeneity of the effect estimates arising from differences in outcome, exposure, and confounder measures. There were insufficient data to tease out independent effects of pollutants by outcome. Unlike outcomes with a long induction period and for which long-term exposures are likely important, perinatal outcomes easily can be studied in prospective studies ARTICLE IN PRESS M. Maisonet et al. / Environmental Research 95 (2004) 106–115 of relatively short length. Prospective studies may provide new information about the meaning of these associations. Such studies could incorporate the use of biomarkers to help elucidate the mechanisms by which ambient pollutants induce an effect and to determine any critical window of exposure. In addition, prospective studies would allow the collection of data on social and community factors not normally collected in vital records and that may modify the association between air pollution and fetal growth (Hughes and Simpson, 1995). References Alderman, B.W., Baron, A.E., Savitz, D.A., 1987. Maternal exposure to neighborhood carbon monoxide and risk of low infant birth weight. Public Health Rep. 102, 410–414. Ananth, C.V., Savitz, D.A., Luther, E.R., 1996. Maternal cigarette smoking as a risk factor for placental abruption, placenta previa, and uterine bleeding in pregnancy. Am. J. Epidemiol. 144, 881–889. Bobak, M., 2000. Outdoor air pollution, low birth weight, and prematurity. Environ. Health Perspect. 108, 173–176. Bobak, M., Leon, D.A., 1999. Pregnancy outcomes and outdoor air pollution: an ecological study in districts of the Czech Republic 1986–8. Occup. Environ. Med. 56, 539–543. Dejmek, J., Selevan, S.G., Benes, I., Solansky, I., Sram, R.J., 1999. Fetal growth and maternal exposure to particulate matter during pregnancy. Environ. Health Perspect. 107, 475–480. Ha, E.H., Hong, Y.C., Lee, B.E., Woo, B.H., Schwartz, J., Christiani, D.C., 2001. Is air pollution a risk factor for low birth weight in Seoul? Epidemiology 12, 643–648. Hughes, D., Simpsom, L., 1995. The role of social change in preventing low birth weight. Future Child. 5, 87–102. Kline, J., 1989. Conception to Birth: Epidemiology of Prenatal Development. Oxford University Press, New York. Landgren, O., 1996. Environmental pollution and delivery outcome in Southern Sweden: a study with central registries. Acta Paediatr. 85, 1361–1364. 115 Lieberman, E., Gremy, I., Lang, J.M., Cohen, A.P., 1994. Low birthweight at term and the timing of fetal exposure to maternal smoking. Am. J. Public Health 84, 1127–1131. Lockwood, C.J., 1994. Recent advances in elucidating the pathogenesis of preterm delivery, the detection of patients at risk, and preventative therapies. Curr. Opin. Obstet. Gynecol. 6, 7–18. Maisonet, M., Bush, T.J., Correa, A., Jaakkola, J.K., 2001. Relation between ambient air pollution and low birth weight in the northeastern United States. Environ. Health Perspect. 109, 351–356. McDonald, A.D., Armstrong, B.G., Sloan, M., 1992. Cigarette, alcohol, and coffee consumption and prematurity. Am. J. Public Health 82, 87–90. Perera, F.P., Whyatt, R.M., Jedrychowski, W., Rauh, V., Manchester, D., Santella, R.M., et al., 1998. Recent developments in molecular epidemiology: a study of the effects of environmental polycyclic aromatic hydrocarbons on birth outcomes in Poland. Am. J. Epidemiol. 147, 309–314. Ritz, B., Yu, F., 1999. The effect of ambient carbon monoxide on low birth weight among children born in Southern California between 1989 and 1993. Environ. Health Perspect. 107, 17–25. Ritz, B., Yu, F., Chapa, G., Fruin, S., 2000. Effect of air pollution on preterm birth among children born in Southern California between 1989 and 1993. Epidemiology 11, 502–511. Rogers, J.F., Thompson, S.J., Addy, C.L., McKeown, R.E., Cowen, D.J., Decoufle, P., 2000. Association of very low birth weight with exposures to environmental sulfur dioxide and total suspended particulates. Am. J. Epidemiol. 151, 602–613. Seaton, A., MacNee, W., Donaldson, K., Godden, D., 1995. Particulate air pollution and acute health effects. Lancet 345, 176–178. Wang, X., Ding, H., Ryan, L., Xu, X., 1997. Association between air pollution and low birth weight: a community-based study. Environ. Health Perspect. 105, 514–520. Williams, M.A., Mittendorf, R., Stubblefield, P.G., Lieberman, E., Schoenbaum, S.C., Monson, R.R., 1992. Cigarettes, coffee, and preterm premature rupture of the membranes. Am. J. Epidemiol. 135, 895–903. Xu, X., Ding, H., Wang, X., 1995. Acute effects of total suspended particles and sulfur dioxides on preterm delivery: a communitybased cohort study. Arch. Environ. Health 50, 407–415.