The Role of Environmental Toxicants in Preterm Birth
Few environmental pollutants have been investigated for their potential to increase the risk for preterm birth. Among those pollutants that have been studied, however, only limited information is available for most of them. Because of this general lack of information, the potential contribution of environmental chemical pollutants to preterm birth is poorly understood. Possible exceptions are lead and environmental tobacco smoke, for which the weight of evidence suggests that maternal exposure to these pollutants increases the risk for preterm birth. In addition, a number of epidemiological studies have found significant relationships between exposures to air pollution and preterm birth, particularly for sulfur dioxide and particulates, suggesting that exposure to these air pollutants may increase a woman’s risk for preterm birth. Studies to date suggest that agricultural chemicals deserve greater attention as potential risk factors for preterm birth. Other studies suggest that follow-up investigations that examine exposures to nitrates and arsenic in drinking water are warranted. In addition, despite persistent racial-ethnic disparities in the rates of preterm birth and increased awareness of racial-ethnic disparities in the levels of exposure to environmental toxicants, few studies have considered the interactions among race-ethnicity, environmental chemical exposure, and preterm birth. Nevertheless, the vast numbers of pollutants to which a woman may be exposed have never been consid-
ered in an investigation of preterm birth. Because of the large number of potential exposures, an efficient and effective research program that will serve public health needs is required.
Various environmental chemical pollutants are recognized as risk factors for numerous diseases and pathophysiological responses, and with the recognition that these pollutants pose a risk, policies have been instituted to protect the public health. Lead is perhaps the most renowned toxicant for which such actions have been taken, with national policies and programs established to protect children from neurotoxicity by removing lead from gasoline and paint. However, the potential risk for preterm birth as a result of exposure to environmental pollutants is poorly understood. This lack of knowledge presents a potentially significant shortcoming in the design of public health preventive strategies.
The present discussion specifically considers epidemiological studies that have analyzed the associations between exposure to environmental chemical pollutants and preterm birth. For this review, preterm birth is defined as the delivery of a live infant before 37 completed weeks of gestation, unless an alternative definition was used by a particular study. Analyses and findings of associations between environmental chemical exposures and low birth weight are not discussed. Because differences in mean gestational age at birth may or may not be relevant for preterm birth, depending on whether or not the distribution is affected at the lower gestational ages, studies in which gestational age was used as a continuous variable are discussed only as they contribute to the understanding of preterm birth.
Studies for consideration were identified on the basis of a PubMed Boolean search that crossed various terms for preterm birth (e.g., “premature birth,” “prematurity,” and “preterm delivery”) or the term “birth outcomes” with general terms for toxicants (e.g., “air pollutants,” “water pollutants,” and “pesticides”) and terms for selected specific pollutants (e.g., “dioxin,” “polychlorinated biphenyls,” and “DDT” [dichlorodiphenyltrichloroethane]). If the results of the studies evaluated were adjusted for confounders, only the adjusted statistics were considered in the final review. If adjustments were not made for potential confounders, the crude odds ratio (OR), crude relative risk (RR; risk ratio), or the most relevant statistical result was considered. The results were considered significant if they achieved statistical significance in the particular study and the confidence interval of the statistical estimate of adjusted OR or RR did not include unity (1.0). Although an attempt has been made to be thorough, some studies in this area may have been overlooked.
EXPOSURE ASSESSMENT CHALLENGES
A variety of approaches have been used to estimate exposures and to investigate the associations between environmental chemical exposures and preterm birth. A common approach has been to use employment or location of residence as a proxy for exposure, such as working with pesticides, proximity to a pollution source, or residence in a polluted locality. In those studies, exposure is typically estimated from measurement of the levels of contamination of common environmental media, such as drinking water sources and ambient air. However, such studies usually lack information about individual exposures and confounders, among other limitations.
Different approaches have been used to obtain information on individual levels of exposure. Perhaps the most conventional approach is to obtain self-reported data on exposures collected by survey or interview, but this method has the potential for reporting bias. Alternatively, measurements of the levels of chemical exposure have been used. These typically involve measurement of the levels of selected pollutants or their metabolites in body fluids or tissues, such as maternal blood, umbilical cord blood, and the placenta. However, the tissue or body fluid selected for toxicant level analysis has implications for the study, depending on the distribution of the toxicant in the body during pregnancy and the expected tissue or body fluid target for the action of the toxicant. Information on individual exposures provides potential advantages, such as controlling or adjusting for confounding variables and minimizing exposure misclassification.
The timing of the assessment may also influence the exposure measure. For example, if the toxicant concentration in the tissue changes over the course of gestation, the toxicant concentrations at the time of birth for preterm births may differ from those for term births simply as a function of gestational age. Similarly, if self-reporting or geographical-ecological assessments are used, the results may vary depending on whether the exposure is assessed for the last month of pregnancy, the entire pregnancy, or even for a period before the pregnancy. Co-exposures are common for environmental pollutants and pose another exposure assessment challenge. For example, cigarette smoke contains thousands of chemicals, many with known toxicity. Sulfur dioxide and particulates are typical co-pollutants of air, and many of the fat soluble toxicants are co-pollutants of food. Few studies have addressed potential impacts of co-pollutants in studies of preterm birth.
A major challenge is the assessment of cumulative exposures. Some pollutants are stored in the body, such as lead in bone or the pesticide DDT in fat. Metabolic changes of pregnancy cause increased metabolism of bone and fat tissues, such that toxicants are released from these sites to the blood and general circulation. Moreover, the impact of cumulative exposure may
be important for preterm birth, but has been largely unexplored. However, cumulative exposure assessments face many challenges, in part because individuals are often unaware of exposure levels. Consequently, the development of biological markers for estimating cumulative exposure (e.g., k-x-ray fluorescence measurements of lead in bones) is needed for studies to assess impacts of cumulative exposures.
A significant and relatively recent effort has been made to link environmental air pollution exposures with preterm birth. Recent reviews examined a subset of articles on this topic in detail and drew different conclusions (Maisonet et al., 2004; Sram et al., 2005). Sram et al. (2005) reported that the evidence was insufficient to derive conclusions about an association between air pollution exposure and preterm birth and that further studies were justified. In contrast, Maisonet et al. (2004) reported that air pollution has an apparent effect on the rates of preterm birth, although the effects are relatively small, and that the conclusions were limited by the small number of studies and differences in the measurements of outcome, exposure, and confounders. The reviews by Sram et al. (2005) and Maisonet et al. (2004) included only 4 studies each, however, whereas the committee located 21 studies in the scientific literature for the present discussion on this topic. This section provides a brief review of the approaches, major findings, and challenges that is more inclusive than those provided in the reviews by Sram et al. (2005) and Maisonet et al. (2004). Moreover, some exposures that may have had an air pollution component but that fit better in other categories, such as lead exposure, are discussed elsewhere in this chapter in the discussion of the specific toxicant.
Different approaches have been used to assess exposure to air pollution. The most common approach has been to use residential proximity to air pollution sources, such as highways or oil refineries, as a proxy for exposure. An alternative exposure assessment approach relied on fixed air-monitoring stations to estimate the levels of air pollution exposure in the study population. An advantage of the latter approach is that specific pollutants were analyzed for their associations with preterm birth. However, estimates of the health effects of air pollutants tend to be smaller when exposure is assessed by using fixed-site air-monitoring station data rather than individual exposure data obtained by sampling individual women (Navidi and Lurmann, 1995). Moreover, the proximity of an individual’s residence to a monitoring station was shown to influence the statistical association with preterm birth, further illustrating the heterogeneity of exposure as a limitation of this approach (Wilhelm and Ritz, 2005). Additionally, studies showed that pregnant women who were socioeconomically dis-
advantaged and of a nonwhite race were more likely to live in neighborhoods with more air pollution (Ponce et al., 2005; Woodruff et al., 2003). Because socioeconomic disadvantage and African American race are risk factors for preterm birth (see Chapter 4), geographical or ecological assessments of exposure to pollutants may be confounded by socioeconomic condition and race.
A variety of comparison strategies have been used. Many studies compared women who resided in a more polluted community with women who resided in a less polluted community. Some studies that used stationary air-monitoring station data to estimate exposures evaluated exposure-response relationships by analyzing the trends in preterm birth rates with respect to the exposure level on the basis of the woman’s residence. Another approach has been to study the same population over time, because air pollutant levels vary significantly due to short-term fluctuations and the seasons (Sagiv et al., 2005; Sram et al., 1996; Wilhelm and Ritz, 2005). Moreover, the air concentrations of certain pollutants are known to correlate with one another. Because of this potential confounding, some studies that monitored exposure to multiple air pollutants also analyzed or adjusted for copollutant effects (Sagiv et al., 2005; Xu et al., 1995).
Sulfur Dioxide and Particulates
The most consistent relationships between specific air pollutants and preterm birth have been reported for sulfur dioxide and particulates. Of particular relevance to the United States is a study of births in Vancouver, British Columbia, Canada, conducted between 1985 and 1998, which reported an association between sulfur dioxide air pollution and preterm birth with an adjusted OR of 1.09 (95% confidence interval [CI] = 1.01–1.19) per 5.0 parts per billion (ppb) (14.3 micrograms per cubic meter [μg/m3]) of sulfur dioxide (Liu et al., 2003). The mean daily concentration of sulfur dioxide during the study period was 4.9 ppb, with a maximum peak concentration (over 1 hour) of 128.5 ppb. Studies of women in the Czech Republic and Beijing, China, who were exposed to sulfur dioxide pollutant concentrations higher than usual for those in the United States reported concentration-dependent relationships between ambient air sulfur dioxide concentrations and preterm birth, with adjusted ORs of 1.27 (95% CI 1.16– 1.39) per 50-μg/m3 increase in the sulfur dioxide concentration (Bobak, 2000) and 1.21 (95% CI 1.01–1.45) for each ln natural log unit μg/m3 increase in the sulfur dioxide concentration (Xu et al., 1995). The rates of preterm births also increased among women living in the vicinity of a coalburning power plant in Croatia during a period of high-intensity power plant operation with higher sulfur dioxide emissions compared with those
during a period of lower-intensity power plant operation (RR = 1.76; p = 0.026) (Mohorovic, 2004).
Similar to sulfur dioxide, associations between exposure to airborne particulates and preterm birth have been reported. Airborne particulates have been measured as total suspended particulates and particulates equal to or less than 10 micrometers (μm) in diameter (PM10). In a study of women in Southern California, PM10 exposure in the 6-week period before birth increased the risk for preterm birth by vaginal delivery (adjusted RR 1.19 per 50 μg of total suspended particulates/m3; 95% CI 1.10–1.40) and preterm birth by delivery by cesarean section (adjusted RR 1.35 per 50 μg of total suspended particulates/m3; 95% CI 1.06–1.69) (Ritz et al., 2000). Concentration-dependent responses were observed in populations of women in the Czech Republic (adjusted OR 1.18; 95% CI 1.05–1.31 per 50 μg of total suspended particulates/m3) (Bobak, 2000) and Beijing (adjusted OR 1.10 μg/m3; 95% CI 1.01–1.20 per 100 μg of total suspended particulates/ m3) (Xu et al., 1995) who were exposed to concentrations of particulate air pollution higher than those usually found in the United States.
In contrast, a time-series study of four Pennsylvania counties found suggestive, but not significant, associations between exposure to sulfur dioxide or PM10 and preterm birth (Sagiv et al., 2005). Likewise, the linkage of medical registries with air pollution data on sulfur dioxide, hydrocarbons, and nitrogen oxides in southern Sweden found no differences in the rates of preterm birth in comparisons of the municipalities with pollutant levels above and below the mean level and when the rate of preterm birth in the municipality with the highest pollutant level was compared with the rates in all other municipalities (Landgren, 1996).
Carbon Monoxide, Nitrogen Oxides, and Ozone
The relationship of the rates of preterm birth to exposure to carbon monoxide, nitrogen oxides, and ozone pollution is less certain. Exposure to carbon monoxide in ambient air in the last month of pregnancy was associated with preterm birth in a population of women in Vancouver, British Columbia, Canada (Liu et al., 2003), with an adjusted OR of 1.08 (95% CI 1.01–1.15) for each 1.0-part-per-million increase in the carbon monoxide level. In contrast, the relationship between carbon monoxide pollutant levels and preterm birth was inconsistent in a study of pregnancies in Southern California: exposure to carbon monoxide in the 6-week period before birth was associated with increased rates of preterm birth for inland regions (RR 1.12; 95% CI 1.04–1.21) but with decreased rates of preterm births for coastal regions (RR 0.77; 95% CI 0.64–0.91) (Ritz et al., 2000).
Disparate results have also been reported for nitrogen oxides. Maroziene and Grazuleviciene (2002) reported increased rates of preterm
birth among Lithuanian women in association with exposure to nitrogen dioxide in ambient air (adjusted ORs for medium and high tertile exposures to nitrogen dioxide were 1.14 [95% CI 0.77–1.68] and 1.68 [95% CI 1.15– 2.46], respectively). However, other studies failed to find a significant association between the levels of nitrogen oxide exposure and the rates of preterm birth (Bobak, 2000; Liu et al., 2003; Ritz et al., 2000).
Likewise, ozone was not significantly associated with preterm birth in two studies (Liu et al., 2003; Ritz et al., 2000).
Location of Residence
Several studies found that women who live in an area with high levels of air pollution are more likely to deliver preterm than women who live in less polluted areas. Women who live near petroleum refinery plants (Lin et al., 2001; Yang et al., 2004a), petrochemical industrial complexes (Yang et al., 2002a,b), or industrial districts with increased levels of emission of air contaminants from multiple sources (including petrochemical, petroleum, steel, and shipbuilding industries) (Tsai et al., 2003) had an increased risk for delivery of preterm infants, with ORs ranging from 1.03 to 1.41. Women living in a mining district of the Czech Republic had nearly twice the prevalence of preterm birth compared with that of women living in a less polluted district (p < 0.01), although that study was confounded by cigarette smoking and racial differences that stratified with the districts (Dejmek et al., 1996; Sram et al., 1996). Residential proximity to highways was also found to be related to the risk for preterm birth, with ORs or RRs ranging from 1.08 to 1.30 (Ponce et al., 2005; Wilhelm and Ritz, 2003, 2005; Yang et al., 2003a).
Exposure by Stage of Gestation
Several studies specifically analyzed whether a stage of gestation for exposure to air pollutants is the most strongly associated with preterm birth. The results have been inconsistent, with some studies reporting significant associations early or late in gestation. Exposure to sulfur dioxide (Bobak, 2000; Mohorovic, 2004), total suspended particulates (Bobak, 2000), and nitrogen dioxide (Maroziene and Grazuleviciene, 2002) in the first trimester, but not the second or the third trimester, of pregnancy was associated with preterm birth. Among women living in a mining district of the Czech Republic who smoked cigarettes during pregnancy, there was an increased prevalence of preterm births for pregnancies conceived in the winter compared with pregnancies conceived in summer. Because winter in this study was characterized by weather inversions that resulted in unusually high concentrations of fine particles dominated by acidic sulfates, genotoxic or-
ganic compounds, and toxic trace elements (Dejmek et al., 1996; Sram et al., 1996), a relationship between conception, exposure and preterm birth is suggested..
In contrast, other studies have reported associations between preterm birth and exposures to toxicants later in pregnancy. Specifically, Liu et al. (2003) found that the rates of preterm birth increased with exposure to sulfur dioxide and carbon monoxide during the last but not the first month of pregnancy. In addition, the most significant elevation in the rates of preterm birth among women who lived near roadways in Los Angeles County with heavy traffic occurred for women whose third trimester fell during the fall and winter months, when the level of traffic-related air pollution was the highest (Wilhelm and Ritz, 2003). In one study, preterm birth was associated with exposure to particulate air pollution (PM10) in the last 6 weeks but not the first month of pregnancy (Ritz et al., 2000). Bobak (2000) found no differences in the odds of preterm birth among the trimesters of gestation (adjusted OR range = 1.24 to 1.27 per 50 μg of increase in the sulfur dioxide concentration/m3). A detailed time-series analysis of exposure to sulfur dioxide and fine particulates (PM10) found suggestive, although not significant, associations of preterm birth with the levels of PM10 and sulfur dioxide exposures in the last 6 weeks and the last 1 week of gestation (Sagiv et al., 2005).
Significance of Overall Findings for Air Pollution
It should be noted that the adjusted ORs or RRs in all of the studies were relatively low (less than 1.68), with many estimates being much lower. The relatively small increased risk of preterm birth reflected by the relatively low ORs or RRs may represent a truly marginal increased risk or may be a consequence of the exposure assessment complexities discussed earlier. In addition, several of the studies analyzed multiple air pollutants, and the tendency to focus on positive findings should be balanced by consideration of the entire context of the study. Nonetheless, given the number of studies that have found significant relationships between exposures to air pollution and preterm birth, the epidemiological findings suggest that air pollution contributes to a woman’s risk for preterm birth. Moreover, several studies reported exposure-response relationships of significance, particularly for sulfur dioxide and particulates (Bobak, 2000; Liu et al., 2003; Maroziene and Grazuleviciene, 2002; Ritz et al., 2000; Sagiv et al., 2005; Xu et al., 1995). Inconsistent findings that were regionally dependent were reported for carbon monoxide, and contradictory results were reported for nitrous oxides. Consequently, conclusions about the risk for preterm birth from ambient air exposures to carbon monoxide and nitrous oxides cannot be made at this time. Moreover, the two studies that evaluated ozone found
that ozone exposure has no significant association with preterm birth, suggesting that ozone may not be a significant contributor to preterm birth. Likewise, the contradictory findings do not allow conclusions regarding a critical gestational stage for exposure to air pollutants and an increased risk of preterm birth.
Finally, the analysis by Xu et al. (1995) deserves further comment. Those investigators found that exposure to sulfur dioxide and total suspended particles modified the distribution of gestational age at birth, such that the largest effects were observed with exposures at the youngest gestational ages. Those researchers calculated an attributable risk (the proportion of cases of preterm delivery in the sample attributable to air pollution) of 33.4 percent. This high attributable risk indicates that exposure to elevated levels of sulfur dioxide and particulate air pollution poses a significant public health concern for pregnant women. Although the concentrations of sulfur dioxide and particulate air pollution that Xu and colleagues (1995) measured in Beijing air are not anticipated in the United States, the study nonetheless illustrates the potential and preventable impacts that air pollutants can have on preterm birth.
A variety of agricultural chemicals are manufactured and widely applied in the United States and worldwide to control pests and enhance agricultural productivity. Human exposures may result from occupational manufacture or use or may be indirect as a result of contamination of environmental media, such as water, air, and food. Among the agricultural chemicals, pesticides have been the most intensively studied for their association with preterm birth. Some of the most notorious pesticides, such as the insecticide DDT, persist in the environment, are poorly excreted, and (biomagnify) in the food chain, thereby increasing potential for human exposure.
Exposure to agricultural chemicals has been assessed by a variety of methods in epidemiological studies of preterm birth. The most common approach has been measurement of the pollutant concentrations in maternal blood, umbilical cord blood, or neonatal blood. Other approaches have used residence on a farm, survey assessment of occupational exposure, and retrospective surveys of exposures of the mother.
DDT has been examined more than any other pesticide in epidemiological studies of preterm birth. DDT is an environmentally persistent insecticide that biomagnifies in the food chain, with known disastrous reproductive consequences for certain wildlife. Although the use of DDT in the United States has been discontinued since 1972, it continues to be applied worldwide to kill mosquitoes for the control of malaria, and environmental
residues persist because of its poor ability to degrade biologically in the environment. DDT is applied as a technical grade of isomers, of which p,p′-DDT is the most abundant. Much of the interest in the o,p′-DDT isomer has been due to its affinity for estrogen receptor alpha. Human excretion of DDT is nominal, although it is metabolized to its persistent metabolite, (p,p′-DDE), which is also poorly excreted. Consequently, studies of DDT have monitored for p,p′-DDT, o,p′-DDT, p,p′-DDE, and, less frequently, various other isomers of DDT. A recent publication includes a comparison table of published reports of studies that analyzed for an association between DDT exposure and preterm birth (Farhang et al., 2005).
Two studies examined the associations between DDT exposure and preterm birth for similar exposure measures over a similar time period and used logistic regression analysis that adjusted for similar confounding variables. The most comprehensive assessment of the association between exposure to DDT and preterm birth included 2,380 women (361 of whom delivered prematurely) who had participated in the U.S. Collaborative Perinatal Project between 1959 and 1965 (Longnecker et al., 2001). By using a logistic regression analysis of third-trimester maternal serum DDE concentrations, Longnecker and colleagues (2001) found that the odds of preterm birth increased steadily and significantly with increasing concentrations of serum DDE (p < 0.0001), with adjusted ORs of 2.5 (95% CI 1.5–4.2) when levels in serum were 45 to 59 μg of DDE/liter and 3.1 (95% CI 1.8– 5.4) when levels in serum were ≥60 μg of DDE/liter. The same study failed to find a significant association between preterm birth and either third-trimester maternal serum DDT concentrations or the ratios of maternal serum DDT levels to serum DDE levels.
In contrast, a logistic regression analysis failed to detect an association between preterm birth and maternal serum concentrations of DDE or DDT or the DDT:DDE ratio among 420 women who had participated in the Child Health and Development Studies of the San Francisco Bay Area from 1959 to 1967 (Farhang et al., 2005). However, that study had significant limitations, such as the inclusion of only 33 preterm births; the inclusion of only infants who had survived until 2 years of age; and the use of maternal blood samples, the majority of which were taken at an unspecified time in the postpartum period, with the untested assumption that the postpartum and pregnancy serum DDT and DDE concentrations would be interchangeable.
Other studies used a case-control approach and relatively small sample sizes (4 to 24 cases of preterm birth). Those studies compared the tissue DDT concentrations and the concentrations of other pesticides in women who delivered preterm with those in women who delivered at term. Many of these studies analyzed for multiple persistent toxicants and emphasized
the positive results. Nonetheless, the findings deserve consideration, albeit with some attention to the context of the study.
Berkowitz and colleagues (1996) reported the only case-control study that matched for potential confounding variables (maternal age, race, and prepregnancy body mass index) and analyzed other potential confounders in a group of New York women who gave birth between 1990 and 1993. They found no significant differences in first-trimester maternal serum DDE levels for women who delivered preterm (n = 20) compared with those for women who delivered at term (n = 20).
Three case-control studies that did not control for potentially confounding variables reported that increased tissue concentrations of several pesticides were associated with preterm birth. In two studies of pregnant women from Lucknow, India, Saxena et al. (1980, 1981) found elevated concentrations of three major isomers of DDT (including DDE), as well as the pesticides lindane, aldrin, and hexachlorobenzene, in the blood and placentas of women who went into preterm labor (defined as labor during 12 to 32 weeks of gestation in the first study; preterm labor was not defined in the second study) compared with those in the blood and placentas of women who delivered at term (p < 0.001). Likewise, a study of Israeli women who delivered preterm reported elevated maternal blood concentrations of the pesticides lindane, dieldrin, heptachlor, and several isomers of DDT, including DDE, at the time of delivery compared with the concentrations in women who delivered at term (p < 0.02) (Wassermann et al., 1982).
In a study of Brazilian births, elevated concentrations of DDT and DDE were observed in the umbilical cord blood of preterm infants compared with those in term infants, even though there were no differences in maternal blood DDT concentrations (Procianoy and Schvartsman, 1981). A study of births in Flix, Spain, of women exposed to extremely high levels of atmospheric pollution from an electrochemical plant found umbilical cord blood DDE concentrations that were three times higher for preterm births (n = 4) than for term births (n = 66) (p < 0.05) (Ribas-Fito et al., 2002), but no significant differences in the concentrations of the pesticide β-hexachloro-cyclohexane were detected.
Although they do not tend to persist in the environment, organophosphate insecticides comprise another category of pesticides with potential for human exposure because of their widespread use. By using inhibition of cholinesterase enzyme activity in umbilical cord blood as a biomarker of exposure to organophosphates, Eskenazi et al. (2004) reported an increased odds for preterm birth in association with organophosphate pesticide exposure (adjusted OR 2.3; 95% CI 1.1–4.8; p = 0.02). Although the concentrations of metabolites of organophosphate pesticides in maternal urine were not associated with preterm birth in the latter study, the urine samples were
collected before parturition, whereas the umbilical cord blood was collected at the time of delivery. In a population of primarily Hispanic, low-income women assessed for exposure to organophosphate pesticides through analysis of maternal blood cholinesterase activity and exposure history by self-report, no differences in the risk of preterm birth were observed between exposed and unexposed women (Willis et al., 1993).
By using employment in the floriculture industry as a proxy for exposure to pesticides and self-reported data on pregnancy outcomes, Restrepo et al. (1990) compared the prevalence of preterm birth before and after employment for floriculture industry workers in Colombia. They reported an increased odds of preterm birth after working in floriculture for female workers (OR 1.86; 95% CI 1.59–2.17; p < 0.01) and the wives of male workers (OR 2.75; 95% CI 2.01–3.76; p < 0.01).
Paternal exposure, but not maternal exposure, to pesticides was associated with preterm birth in a study of U.S. Navy women that used the mother’s retrospective self-report to assess the father’s occupational exposure to pesticides (Hourani and Hilton, 2000). Exposure to the widely applied herbicide atrazine via drinking water was not significantly associated with preterm birth in logistic regression analyses (Villanueva et al., 2005).
Application of inorganic and organic fertilizer (e.g., manure) to crops is a potential source of nitrate contamination of drinking water supplies. A population-based case-control study with 321 cases found that women in Prince Edward Island, Canada, who lived in regions on the island with higher concentrations of nitrates in the drinking water were more likely than women who lived in the region on the island with the lowest nitrate water levels to deliver the infants preterm (Bukowski et al., 2001a). This association of nitrates in drinking water with preterm birth was exposure dependent, with adjusted ORs increasing from 1.82 (95% CI 1.23–2.69) at the median nitrate concentration of 3.1 milligrams per liter (mg/liter) to ORs of 2.33 (95% CI 1.46–3.68) and 2.37 (95% CI 1.07–4.08) at median nitrate concentrations of 4.3 and 5.5 mg/liter, respectively (Bukowski et al., 2001a). In contrast, a study of 486 West African infants found no association between preterm birth and living in a region with very high nitrate levels in the drinking water, when regions with low and high nitrate were compared according to whether the well water nitrate concentration was below or above 20 mg/liter (Super et al., 1981).
Another study compared Norwegian farmers with a nonfarmer reference group (n = 192,417 births) and found no significant relationship between owning a farm and the risk for preterm birth if all births before 37 completed weeks of gestation were included in the analysis (Kristensen et al., 1997). However, among farmers there was an increase of extremely early, midgestation preterm births (between 21 and 24 weeks of gestation), with an adjusted OR of 1.54 (95% CI 1.23–1.93) for singleton births and a
much higher adjusted OR of 3.75 (95% CI 1.72–8.20) for multiple gestations. Moreover, that study found significant associations with grain farming in particular (OR 1.58; 95% CI 1.19–2.09) and also found a pattern of association with the season of the year and a poor grain harvest quality, which led the authors to speculate that mycotoxins produced by molds may induce midpregnancy preterm labor, with the strongest effect on multiplefetus pregnancies.
Although the epidemiological findings do not present a consensus on whether exposures to particular agricultural chemicals increase the risk for preterm birth, the studies that have been performed to date suggest that agricultural chemicals deserve greater attention as potential risk factors for preterm birth. In particular, the report by Longnecker et al. (2001) provides the strongest evidence for an association of DDT exposure with preterm birth because of the large sample size, the significant exposure-response relationship, and adjustment for confounders. Although DDT exposure may continue to be significant elsewhere in the world, it should be noted that the latter study used blood samples collected from women in the United States between 1959 and 1967, when exposures to DDT were substantially greater in the United States than they are today. Other studies of DDT and its association with preterm birth are compromised by small sample sizes, a lack of subject demographic information, or a lack of consideration for confounders.
Early reports that associated preterm birth with elevated tissue levels of other persistent organochlorine pesticides (lindane, aldrin, dieldrin, heptachlor, and hexachlorobenzene) have not been pursued. The report that preterm birth was significantly associated with a biomarker in umbilical cord blood for exposure to organophosphate pesticides (Eskenazi et al., 2004) requires follow-up. Given the potential for exposure from contaminated drinking water, further examination of the risk for preterm birth from nitrates is also warranted. Finally, the potential for mycotoxins to contribute to early preterm births is intriguing (Kristensen et al., 1997) and deserves further analysis by the use of individual-specific exposure assessments.
Polychlorinated biphenyls (PCBs) are industrial chemicals that were manufactured and used commercially as mixtures of congeners, with each congener distinguished by the number and pattern of chlorination of the biphenyl rings. Certain PCB congeners persist in the environment, are poorly metabolized, and biomagnify in the food chain. These characteristics result in increased opportunities for human exposure. The production and commercial use of PCBs were discontinued in most countries in the late 1970s,
although there is evidence that production continued in some countries until recently.
An early study reported that the concentration of total PCBs in the blood of 17 women who delivered prematurely was 3.7-fold higher than that observed in 10 women with normal third-trimester pregnancies (p < 0.025) and found that penta- and hexachlorinated biphenyl congener levels were specifically elevated in the maternal blood of women who delivered preterm (Wassermann et al., 1982). However, no demographic information on the women in the latter case-control study was available. Longnecker et al. (2005) failed to find a significant association between PCBs and preterm birth in a logistic regression analysis of 1,034 historical maternal blood samples collected between 1959 and 1965 that included potential confounding factors. Likewise, a case-control study failed to find differences in maternal serum levels of PCBs (total or individual congeners) between New York City women who delivered preterm (n = 20) or full term (n = 20) between 1990 and 1993 in a study that matched the women for potential confounding variables (maternal age, race, and prepregnancy body mass index) and that analyzed for other potential confounders (Berkowitz et al., 1996). Furthermore, there were no significant differences between PCB concentrations in umbilical cord serum from preterm births compared with those from full-term births in a Spanish population exposed to extremely high levels of air pollution (Ribas-Fito et al., 2002).
Because PCBs biomagnify in the food chain, two studies examined relationships between the consumption of Great Lakes fish, PCB exposure and preterm birth. In a study conducted in Michigan, women who reported eating the equivalent of about two salmon or lake trout meals per month had decreased gestational lengths (an average of 4.9 days shorter) compared with those of women who ate less Great Lakes fish, and infants with higher concentrations of PCBs in the umbilical cord serum were born an average of 8.8 days earlier than infants with lower cord serum PCB concentrations (Fein et al., 1984). In contrast, a later study conducted in Green Bay, Wisconsin, found no significant associations between maternal reports of Lake Michigan fish consumption or maternal serum concentrations of PCBs and any adverse reproductive outcome (Dar et al., 1992).
In an analysis of a population of women who worked in two capacitor manufacturing facilities in upstate New York, women with higher levels of occupational exposure to PCBs had significantly decreased gestational durations (an average of 6.6 days shorter) (Taylor et al., 1984). A follow-up study of the women in the capacitor manufacturing facilities found that the differences in infant birth weights between women exposed to high levels of PCBs and women with low levels of exposure to PCBs was explained, at least in part, by differences in gestational duration (Taylor et al., 1989).
The largest and most rigorous studies (with respect to control for con-
founding factors) failed to establish a significant association between exposure to PCBs and preterm birth. Because of the epidemiological findings and the fact that environmental contamination from PCBs is declining in environmental media in most regions of the world, the studies do not support the possibility that exposure to PCBs is likely a significant risk factor for preterm birth in women currently residing in the United States.
Chlorinated dioxins comprise a class of compounds made up of 75 structurally related congeners that are distinguished by the number and position of chlorine atoms on the dibenzo-p-dioxin molecule. The most toxic and well known chlorinated dioxin is 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD), commonly referred to as dioxin. Although TCDD and other dioxins are produced naturally by forest fires or volcanic activity from the incomplete combustion of organic matter, it is the anthropomorphic production as unintended by-products of incineration or combustion that contributes the greatest amount of pollution of chlorinated dioxins in the environment. TCDD was also an unintended by-product in the production of the defoliating herbicide Agent Orange, which was widely applied in southeastern Asia during the Vietnam War.
Several studies have examined whether exposure to dioxin is associated with preterm birth, but the findings are contradictory. Serum TCDD levels of women living in Seveso, Italy, near the site of a major industrial accident that released TCDD into the air, were not significantly associated with an increased odds of preterm delivery (Eskenazi et al., 2003). In a survey study of cancer and reproductive outcomes in the TCDD-contaminated Russian town of Chapvaevsk, the average rate of preterm labor was reported to be significantly higher in women who lived in Chapvaevsk than in women who lived in other towns in the region (Revich et al., 2001). However, preterm birth was not defined, and statistical comparisons were not provided.
In a pilot qualitative study conducted in the year 2000, higher numbers of preterm births were reported among 30 Vietnamese women whose husbands or who themselves were exposed to TCDD through the herbicide Agent Orange (Le and Johansson, 2001). Because TCDD mediates most, if not all, of its toxicity by the binding and activation of the Ah receptor, one study measured the placental concentration of Ah receptor sites, the affinity of TCDD for the Ah receptor, and aryl hydroxylase activity (a cytochrome P450 CYP1A1 enzyme regulated by the Ah receptor) but found no significant differences between normal and preterm births (Okey et al., 1997).
Consequently, despite the general toxicity and notoriety of TCDD, epidemiological studies have failed to find a significant relationship between
maternal exposure to TCDD and preterm birth. Consequently, the epidemiological reports do not provide support the possibility that maternal exposure to TCDD is an important risk factor for preterm birth.
CHLORINATION DISINFECTION BY-PRODUCTS
Because of the widespread use of chlorination as a means to disinfect drinking water supplies, there is considerable interest in the potential health risks from the by-products that are formed because of chlorination disinfection. The principal by-products are trihalomethanes (e.g., chloroform, bromodichloromethane, and dichlorobromomethane), haloacetic acids (e.g., trichloroacetic acid and dichloroacetic acid), and 3-chloro-4-(dichloromethyl)-5-hydroxy-2(5H)-furanone.
Studies of the health effects of chlorination disinfection by-products are complicated by exposure assessment challenges. Specifically, the levels of trihalomethanes increase during the summer and autumn months, when organic material from leaves and other sources increasingly infiltrate surface water supplies and thereby favor the formation of trihalomethanes during chlorination disinfection. Additionally, the concentrations of disinfection by-products vary depending on their location in the distribution system, such that the levels of some by-products (e.g., bromodichloromethane) in the supply system increase over time but the levels of haloacetic acids seem to decrease over time. Moreover, individual behaviors can significantly affect exposure. For example, bathing habits and whether a woman drinks tap water or bottled water can significantly modify individual exposures. Studies that rely on geographical or ecological exposure assessments of community drinking water samples are limited by the inability to account for individual exposure variations.
Epidemiological studies of associations between chlorination disinfection by-products and preterm birth or gestational age published before 2002 have been reviewed in detail by Bove et al. (2002) and Graves et al. (2001). The conclusions of those reviews are summarized here, but the individual studies included in those recent reviews are not discussed in detail. Rather, the reader is referred to these reviews for details on the specific studies.
Bove et al. (2002) reviewed eight studies and Graves et al. (2001) reviewed seven studies that analyzed for associations between preterm birth and contamination of drinking water with chlorination disinfection byproducts. Among the studies reviewed by Bove et al. and Graves et al., only a single study (Yang et al., 2000) found a significantly increased odds for preterm birth (adjusted OR 1.34; 95% CI 1.1–1.6) in a comparison of births in municipalities in Taiwan that chlorinated the drinking water supply with births in municipalities that did not. Consequently, the conclusion of both
reviews was that exposure to disinfection by-products was not likely associated with preterm birth.
Additional studies found no associations between chlorination disinfection or chlorination disinfection by-products and an increased risk for preterm birth, including a study published in 1982 (Tuthill et al., 1982) but not included in the reviews by Bove et al. (2002) and Graves et al. (2001), as well as more recent studies (Aggazzotti et al., 2004; Wright et al., 2003, 2004). In fact, Wright et al. (2004) reported a slight but significantly reduced odds of preterm delivery with maternal exposure to the chlorination disinfection by-products total trihalomethanes, chloroform, and bromodichloromethane.
The only study to identify a positive association between exposure to chlorination disinfection by-products and preterm birth based its exposure assessment on the method used to chlorinate the drinking water supply rather than the amounts of by-products in the water. Consequently, the only positive study is limited by the potential for exposure misclassification. In contrast, many of the studies that reported a lack of a relationship or even a positive relationship between exposure to chlorination disinfection by-products and preterm birth relied on specific measures of exposure. Thus, the general result is that the studies fail to support a relationship between preterm birth and exposure to disinfection by-products.
ENVIRONMENTAL TOBACCO SMOKE
Environmental tobacco smoke constitutes passive exposure to cigarette smoke in the ambient air as opposed to active exposure through cigarette smoking. Risks to preterm birth from active cigarette smoking are covered in Chapter 3. A handful of studies have analyzed for an association between environmental tobacco smoke and preterm birth. Among those studies, maternal self-reporting was the most commonly used exposure assessment measure, although two studies used biomarkers of exposure (the maternal serum concentration of the nicotine metabolite cotinine and the concentration of nicotine in maternal hair). Each of the studies reported a positive association between exposure to environmental tobacco smoke and an increased risk for preterm birth. However, the studies have specific nuances that distinguish the findings.
Among women interviewed a few days after delivery for retrospective self-assessment of exposure to environmental tobacco smoke, daily exposure to environmental tobacco smoke of 7 hours or more was significantly associated with an increased risk for preterm birth (OR 1.86; 95% CI 1.05– 3.45) (Hanke et al., 1999). In another study that relied on maternal self-reporting, exposure to environmental tobacco smoke for 7 hours or more
daily was associated with an increased odds for preterm birth (less than 35 weeks of gestation; adjusted OR 2.4; 95% CI 1.0–5.3) but not with an increased risk for preterm birth before 37 weeks of gestation (Windham et al., 2000). However, the latter study was prospective and assessed exposure based on maternal self-reporting of estimated exposure to environmental tobacco smoke in the 1 week before an interview during early to midpregnancy.
Using maternal self-reporting of exposure to the cigarette smoke of a household member during pregnancy as the index of exposure, Ahluwalia et al. (1997) reported a significantly increased risk for preterm birth for women age 30 years or older (adjusted OR 1.88; 95% CI 1.22–2.88) but not younger women. Also relying on maternal self-reports of exposure, Ahlborg and Bodin (1991) found that exposure to environmental tobacco smoke in the workplace but not in the home was associated with an increased risk of preterm birth. As assessed by questionnaire information in another study, the odds of preterm birth were increased with exposure to environmental tobacco smoke at work only (adjusted OR 2.35; 95% CI 0.50–11.1) and, to a greater extent, with exposure at both work and home (adjusted OR 8.89; 95% CI 1.05–75.3) but not with home exposure only (Jaakkola et al., 2001).
Two studies assessed exposure to environmental tobacco smoke on the basis of a combination of interviews of the mother and biomarkers of exposure. Jaakkola et al. (2001) used the nicotine concentration in hair samples obtained after delivery to assess environmental tobacco smoke exposure and found that there was a concentration-dependent association, with the adjusted OR increasing 1.22 (95% CI 1.07–1.39) with each 1 μg of nicotine/gram of hair. In the high-exposure group in the latter study, the odds of preterm delivery were about sixfold higher in comparison with that for the reference group (adjusted OR 6.12; 95% CI 1.31–28.7) (Jaakkola et al., 2001). A second study relied on the concentrations of the nicotine metabolite cotinine in maternal serum at midpregnancy as an index of exposure to environmental tobacco smoke (Kharrazi et al., 2004). The authors of that study reported an increased odds for preterm birth at the highest quintile level of exposure (0.236 to 10 nanograms [ng] of cotinine/milliliter [ml]) compared with that at the lowest quintile (<0.026 ng/ml) (adjusted OR 1.78; 95% CI 1.01–3.13) in a study of 2,777 births (Kharrazi et al., 2004).
Despite differences in exposure assessment and study design, the six published reports show general agreement that exposure to environmental tobacco smoke is associated with an increased risk for preterm birth. The finding by Ahluwalia et al. (1997) of an interaction between exposure to environmental tobacco smoke and maternal age is particularly interesting and deserves further analysis. The adjusted ORs of the studies on the risk of preterm birth as a result of exposure to environmental tobacco smoke
ranged from 1.2 to 2.4, with the exception of the study by Jaakkola et al. (2001), which found adjusted ORs of 6.12 and 8.89. Exempting the study Jaakkola et al. (2001), the magnitude of the increase in the odds for preterm birth as a result of exposure to environmental tobacco smoke is similar to or somewhat higher than that observed for active cigarette smoking (see Chapter 3). Although speculative at this time, one possible explanation for the latter observation may be that environmental tobacco smoke exposures tend to be more constant and longer in duration whereas active cigarette smoking is more episodic. Regardless, the studies suggest that environmental tobacco smoke deserves greater attention as a risk factor for preterm birth. Overall, the published reports indicate that exposure to environmental tobacco smoke should be considered a potential risk factor for preterm birth.
METALS AND METALLOIDS
Metals and metalloids are elements that are found in nature and that can form a variety of chemical moieties. This review considers all chemical forms of a metal or metalloid evaluated in studies that analyzed for associations between exposures and preterm birth. Furthermore, several of the metallic elements constitute the so-called nutritional micronutrients, and deficiencies of these micronutrients have been associated, in some cases, with an increased risk for preterm birth (see Chapter 3). However, this section of the report deals only with excess environmental exposure to metals and metalloids.
Among the metals and metalloids, lead exposure has been studied the most intensively for an association with preterm birth. A thorough review by Andrews et al. (1994) discusses epidemiological studies published before 1994 that analyzed for associations between lead exposure and preterm birth, as well as other birth outcomes. Although noting the weaknesses and the shortcomings of particular studies, Andrews et al. concluded that an adverse effect of lead exposure on preterm delivery was supported because (1) mean lead levels were elevated in women with preterm births compared with those in women with term births in all studies reviewed, and (2) the prevalence of preterm births increased with an increase in the level of maternal exposure to lead in most of the studies reviewed. The specific studies that Andrews et al. reviewed will not be described here; rather, the reader is referred to that review for a detailed discussion (Andrews et al., 1994).
Similar to the results noted by Andrews et al. (1994), Falcon et al. (2003) found that mean placental lead concentrations were significantly (1.5-fold) higher for 18 women who delivered preterm or who had premature rupture of the membranes than for 71 women who delivered at term (Falcon et al., 2003). The latter study of Spanish women found no signifi-
cant differences in age, parity, smoking habits, or location of residential for the women who delivered term and those who delivered preterm.
Torres-Sanchez et al. (1999) found that mean umbilical cord lead levels were marginally significantly higher (p < 0.051) for preterm births than for term births, but only for primiparous women. However, Torres-Sanchez et al. (1999) also analyzed their data by separate logistic regression analyses for the primiparous and multiparous women using quartile groupings of umbilical cord lead concentrations and adjusting for various known preterm birth risk factors. Using the lowest quartile group (that with a mean umbilical cord blood concentration of < 5.1 μg of lead/deciliter) as the reference group, they found a significantly increased odds ratio for preterm birth for the primiparous women but not the multiparous women at the three higher lead exposure levels, with adjusted ORs ranging from 2.60 to 2.82 (Torres-Sanchez et al., 1999). In contrast, a small case-control study that compared Swedish and Polish women who delivered preterm (n = 17) or at term (n = 13) found no significant differences in maternal blood, myometrium, or placenta concentrations of lead at the time of delivery (Fagher et al., 1993).
Studies of historical occupational records conducted in Norway found that women who worked during pregnancy in jobs classified as having high levels of lead exposure were at increased risk for preterm birth compared with the risk for women who worked at jobs classified as having nominal levels of lead exposure (adjusted OR 1.93; 95% CI 1.09–3.28) (Irgens et al., 1998).
Many of the preceding studies used lead concentrations in blood as a measure of exposure. However, the lead in whole blood may not reflect bioavailable lead, because the majority of lead in blood (more than 95 percent) is bound to red blood cells and is thus not available to cause harm (Manton and Cook, 1984; Smith D et al., 2002).
Contradictory results were obtained in pairs of studies on exposures to cadmium and arsenic in drinking water. Cadmium concentrations at the time of delivery were significantly higher in maternal blood but not in the myometriums or the placentas of women who delivered preterm (n = 13) than in those of women who delivered at term (n = 11) (Fagher et al., 1993). In a study of women (n = 44) living in a cadmium-contaminated region of China in which the median concentrations were used to form the comparison groups, there were no significant differences in preterm birth rates between women with higher cadmium concentrations and women with lower cadmium concentrations in maternal blood, umbilical cord blood, or the placenta (Zhang et al., 2004).
In a study of women in Bangladesh, the rate of preterm births was 2.5-fold higher among women in a village serviced by arsenic-contaminated drinking water (n = 96) than among women matched for age, socioeconomic condition, level of education, and age at marriage who lived in a
village with much lower concentrations of arsenic in the drinking water (≤0.02 mg of arsenic/liter; n = 96) (p < 0.02) (Ahmad et al., 2001). Moreover, the rates of preterm births were significantly higher for those Bangladeshi women who relied on arsenic-contaminated drinking water for more than 15 years than for Bangladeshi women who drank arsenic-contaminated water for less than 15 years (p < 0.02) (Ahmad et al., 2001). In contrast, maternal arsenic exposure from drinking well water was not significantly associated with a risk of preterm delivery in a study that compared women who lived in an arsenic-contaminated area with women who lived in an area without a history of arsenic contamination in Taiwan, with the townships matched for degree of urbanization (Yang et al., 2003b).
No excess numbers of preterm birth were observed in a community in which there was an accidental spill of aluminum sulfate into the drinking water than in neighboring communities (Golding et al., 1991). However, the population size of the exposed pregnant women in the latter study was relatively small (4 preterm births among 88 exposed pregnant women).
Overwhelmingly, lead has been investigated in more studies than any other metal or metalloid for associations with preterm birth or decreased gestation length. Although the evidence is not unanimous, there is sufficient evidence to suggest that maternal exposure to lead results in an increased risk for preterm delivery. The current level of knowledge is inadequate to determine if there may be a risk for paternal exposure to lead for preterm birth. Studies of associations between preterm birth and exposures to other metals, such as aluminum and cadmium, or to the metalloid arsenic are too limited to draw even tentative conclusions at this time.
PATERNAL EXPOSURES TO ENVIRONMENTAL TOXICANTS
A few studies have considered the role of toxicant exposures of the father in preterm birth. Each of these studies involved exposures in the workplace.
Two studies of Norwegian historical occupational records examined the role of paternal occupational exposure and preterm birth. In the first study, paternal employment in the printing industry, which results in increased occupational exposure to lead and solvents, was not associated with preterm birth of less than or equal to 37 completed weeks of gestation but was significantly associated with an increased odds for early preterm birth at between 16 and 27 weeks of gestation (adjusted OR 8.6; 95% CI 2.7– 27.3) (Kristensen et al., 1993). A second study found that fathers who worked at jobs with moderate to low levels of exposure to lead but not with high levels of exposure had slightly decreased odds for fathering a pregnancy that delivered preterm (adjusted OR 0.89; 95% CI 0.86– 0.93) (Irgens et al., 1998).
A study conducted by the U.S. National Institute for Occupational Safety and Health found no relationship between paternal occupational exposure to the dioxin TCDD and preterm birth. That study used a pharmacokinetic model to estimate worker’s serum TCDD concentration at the time of conception (Lawson et al., 2004). Similarly, a study of veterans of Operation Ranch Hand, who were responsible for spraying herbicides during the Vietnam War, failed to find consistent effects of paternal exposure to TCDD (which is present in Agent Orange) on the rates of preterm birth (Michalek et al., 1998). In the latter study, the paternal dioxin level measured in 1987 or 1992 was extrapolated to the time of conception of the child to estimate the level of TCDD exposure.
In summary, the few studies that have assessed paternal toxicant exposure failed to find evidence of an increased risk for preterm birth as a result of paternal occupational exposure to lead or TCDD.
RACIAL DISPARITIES IN ENVIRONMENTAL EXPOSURES
As discussed more extensively elsewhere in this report, preterm births are more prevalent among African American women than among women of other racial-ethnic groups, and this pattern has persisted over the years. The terms “environmental justice” and “environmental racism” describe the disproportionate burden of environmental pollution on poor and minority populations (Brown, 1995; Silbergeld and Patrick, 2005). A recent review by Silbergeld and Patrick (2005) discusses in detail the disproportionate exposures of those populations to environmental pollutants and the effects on birth outcomes. Although the latter review emphasizes birth outcomes other than preterm birth, its discussion includes reports of racial-ethnic differences in environmental exposures that are relevant to preterm birth. Silbergeld and Patrick (2005) concluded that, “exposures to these toxicants may explain part of the socioeconomic disparity that is observed in terms of risks of adverse pregnancy outcomes” (Silbergeld and Patrick, 2005).
Despite the persistent racial-ethnic disparities in the rates of preterm birth and the increased awareness of racial-ethnic disparities in environmental exposures, few studies have considered the interactions among raceethnicity, environmental chemical exposures, and preterm birth. Woodruff et al. (2003) reported increased levels of air pollution in neighborhoods consisting predominantly of minority populations and, after adjusting for maternal risk factors that included race-ethnicity, found a small increase in the odds of preterm delivery (OR 1.05; 95% CI 0.99–1.12) in association with high levels of air pollution.
Other factors that may influence exposure in racially and ethnically distinct patterns include behavioral, cultural, and sociological characteristics and practices. In a case-control study (188 preterm births and 304 nor-
mal births), use of a chemical hair straightener (relaxer) or chemical curl products by African American women just before or during pregnancy had no effect on the risk for preterm birth (Blackmore-Prince et al., 1999).
More research is needed obtain an improved understanding of environmental pollution with respect to race-ethnicity and preterm birth. Because exposure to environmental chemicals may be codependent with race-ethnicity, examination of racial-ethnic differences in environmental exposures during pregnancy may provide new insight into the racial-ethnic disparities in the rates of preterm birth.
There has been little research on mechanisms by which pollutants may stimulate preterm birth. Conceivably, pollutants could increase risk for preterm birth by prematurely activating physiologic mechanisms of parturition or by activating pathologic mechanisms that prematurely initiate parturition. Consequently, biological mechanisms by which pollutants may stimulate preterm birth could include disruption of the endocrine systems that regulate parturition, activation of cell signaling pathways that stimulate uterine muscle contraction, and activation of the inflammatory pathway, among others (see Chapters 6 for discussion of these mechanisms in preterm birth). Moreover, toxicants may be considered as potential tools to uncover previously unknown aspects of parturition, as has been the case for the nervous system.
Mechanisms of direct stimulation of uterine contraction frequency have been studied in rat uteri exposed to DDT isomers and polychlorinated biphenyls (PCBs). These studies suggest that stimulation of uterine contraction frequency by a DDT isomer and a commercial PCB mixture (Aroclor 1254) involves increased intracellular calcium due to activation of voltagesensitive calcium channels in the uterine smooth muscle cells (Bae et al., 1999b; Juberg and Loch-Caruso, 1992; Juberg et al, 1995). Moreover, stimulation of uterine contraction by the PCB mixture is dependent on activation of a calcium-independent phospholipase A2 and increased release of arachidonic acid (Bae et al., 1999a). A PCB congener (PCB 50) was shown to stimulate rat uterine contractions by activating an endometrial calcium-independent phospholipase A2 that was differentially expressed in late gestation (Brant et al., 2006a). Furthermore, Brant and colleagues showed that PCB 50 activation of phospholipase A2 is mediated by p38 mitogen activated protein kinase (MAPK) (Brant and Caruso, 2005). Although increased intracellular calcium of uterine muscle cells, arachidonic acid release, prostaglandin release, and increased uterine contraction frequency are characteristics of parturition, it is not known if these responses measured in rat cells and tissues in vitro have relevance for human preterm birth.
Recent studies suggest that some toxicants may stimulate an intrauterine inflammatory response. Xu et al. (2005) showed that a phthalate (plasticizer and fragrance stabilizer) and its metabolites increase the expression of peroxisome proliferator-activated receptors (PPARs) in a rat placental trophoblast cell line. Latini and colleagues have proposed that induction of an inflammatory response via PPAR could be a mechanism by which phthalates decrease gestation length (Latini et al., 2003, 2005). Brant and colleagues reported that polybrominated diphenyl ethers (used as flame retardants), stimulate the release of pro-inflammatory cytokines from human gestational membranes (Brant et al., 2006b). Although intrauterine inflammation is associated with preterm birth (see Chapter 6), it remains to be shown whether toxicant-induced inflammation is a mechanism for preterm birth in women. Clearly, there is need for improved understanding of toxicant modification of mechanisms of normal and preterm birth.
Few environmental pollutants have been investigated for their potential to increase the risk for preterm birth, and among those pollutants that have been studied, the information available for most of them is limited. Because of this general lack of information, the potential contribution of environmental chemical pollutants to preterm birth is poorly understood. Possible exceptions are lead and environmental tobacco smoke, for which the weight of evidence suggests that maternal exposure to these pollutants increases the risk for preterm birth. In addition, a number of epidemiological studies have found significant relationships between exposures to air pollution and preterm birth, particularly for sulfur dioxide and particulates, suggesting that exposure to these air pollutants may increase a woman’s risk for preterm birth.
Studies to date suggest that exposures to agricultural chemicals deserve greater attention as potential risk factors for preterm birth. In particular, the report by Longnecker et al. (2001) provides the strongest evidence for an association of DDT exposure with preterm birth, although it should be noted that the exposure levels were substantially higher for the samples used in that study compared with the current levels of DDT exposure in the United States. Other studies suggest that follow-up investigations that examine exposures to nitrates and arsenic in drinking water are warranted.
Although particular pollutants are cited here as deserving further scientific inquiry, it should be noted that the vast numbers of pollutants to which a woman may be exposed have never been considered in an investigation of preterm birth. It would be shortsighted to limit future research on the basis of the paucity of information on pollutants for which some information is already available. However, because of the large number of potential expo-
sures, a strategy for an efficient and effective research program that will serve public health needs is required. Such strategies may be based on pathophysiological mechanisms or chemical structure-activity approaches to investigate related toxicants. Alternative strategies might target the most common pollutants identified, for example, pollutants listed in the National Health and Nutrition Examination Survey or Toxics Release Inventory databases. For example, we know from the NHANES study that phthalates are found in urine of most women and are high in women of childbearing age (Silva et al., 2004). New strategies from investigative lines of research involving proteomics, genomics, and metabolomics will likely evolve.
The studies currently available raise important questions. For example, when is the critical window for pollutant exposure during pregnancy? This question has been addressed only recently in investigations of air pollutants, with no clear answer, and has not been examined for other pollutants. It is likely that the critical period would depend, in part, on the pathway through which the environmental pollutant initiates its action. Because pollutants have widely diverse chemical structures and biological activities, there may not be a single best time during gestation to assess exposure for all environmental pollutants. Furthermore, there may not be a single critical exposure period if multiple mechanisms can be activated. For some pollutants, critical exposures may occur before pregnancy or even during prenatal development or well after birth, say, during adolescence. The exposure assessment methodology will therefore require some understanding of the biological plausibility by which the pollutant under study would be hypothesized to affect preterm birth.
Another important question concerns the methodology used to assess exposure. So-called ecological exposure assessments that rely on the measurement of the levels of pollutants in certain geographical areas are limited by the inability to account for the heterogeneity of exposures and individual confounding factors. Methodologies that assess individual exposures through survey or interview instruments are subject to various biases and may suffer from imprecision. However, individual exposure assessments that use biological or chemical markers also present challenges specific to investigations related to preterm birth. Foremost among those challenges is the selection of the appropriate measurement technique and the fluid or tissue to be sampled. For example, several studies found significant associations by using one biomonitoring technique but not another or by measurement of toxicant concentrations in one tissue or body fluid but not another. Moreover, if a pollutant’s concentration in the tissue or fluid sampled changes as a function of gestational age because of the physiological changes that occur during pregnancy, the concentrations in samples taken at the time of delivery may reflect the gestational age rather than actual differences in levels of exposure.
Progress in understanding the biological basis for associations between exposures to environmental pollutants and preterm birth have been limited, in part, by the significant limitations of current laboratory animal models (see Chapter 6). Additionally, few studies have examined mechanisms by which pollutants may stimulate preterm birth. Because pollutants may increase risk for preterm birth by prematurely activating physiologic mechanisms of parturition or by activating pathologic mechanisms that prematurely initiate parturition, toxicant mechanisms need to consider. Consequently, biological mechanisms by which pollutants may stimulate preterm birth could include disruption of the endocrine systems that regulate parturition, activation of cell signaling pathways that stimulate uterine muscle contraction, and activation of the inflammatory pathway, among others. Clearly, there is need for improved understanding of toxicant modification of mechanisms of normal and preterm birth.
Understanding the potential impact of exposure to environmental pollutants on preterm birth is inherently a complex proposition. However, the challenge posed by its complexity should not negate the importance of the task. By and large, women’s exposures to environmental pollutants are unavoidable and unintended. Furthermore, there is the potential for very large numbers of women to be exposed to particular pollutants, such that an increased risk presented by such exposures could have a significant impact on the population as a whole. Public health therefore plays an important role in regulating such exposures. However, regulatory actions to protect the health of pregnant women and their unborn children cannot be made in a vacuum of information.
As the discussion in this chapter indicates, there is a need to increase and improve research in this area to provide the knowledge foundation needed to design effective public health preventive strategies to minimize the risk of preterm birth as a result of environmental exposures.
Finding 8-1: Limited data suggest that some environmental pollutants, such as lead and tobacco smoke, and air pollution may contribute to the risk of preterm birth; but most environmental pollutants have not been investigated. In addition, the interactions between environmental toxicant exposures with other behavioral, psychosocial, and sociodemographic attributes have been understudied.