Birth certificate template

A MAJOR CONCERN of individuals who live near hazardous waste sites is the potential risk of adverse pregnancy outcomes, especially birth defects. Despite these concerns, few investigations of the reproductive effects resulting from residence near hazardous waste sites have been made. One major problem is that investigators have difficulty assembling affected populations sufficiently large to allow evaluation of risk of birth defects. In four studies, investigators evaluated the risk of birth defects among mothers residing in the vicinity of specific hazardous waste sites; their results were inconclusive.[1-4] In three other studies, researchers evaluated the relationship between risk of birth defects and multiple sources of contamination. Shaw[5] combined data on hazardous waste sites and other sources of industrial pollution and showed that women who gave birth to children with heart and other circulatory malformations were more likely than controls to reside in a census tract that included sites with evidence for human exposure. Recently, Croen et al.[6] found elevated (but not statistically significant) increased risks of neural-tube defects and conotruncal heart defects associated with living in a hazardous waste site within 1/4 mile of a definite or potential exposure source. In New York State, Geschwind et al.[7] found a small, but statistically significant, increased risk for malformations in general, as well as for specific anatomical groupings of malformations among mothers living near a hazardous waste site. Geschwind et al.[7] defined exposure to hazardous wastes primarily by residential proximity to

each site (i.e., [is less than] 1 mi) and provided limited evidence about the potential for human exposure.

We designed our study to follow-up the study conducted by Geschwind et al.[7] Instead of evaluating all malformations and all contamination from hazardous waste sites, we evaluated associations between specific organ systems and chemical groups for which Geschwind[7] reported elevated risks and for which other animal and human data suggested possible causal relationships. We specified environmental exposures more accurately and in greater detail. The associations we evaluated included central nervous system (CNS) defects and solvents, CNS defects and metals, musculoskeletal (MUS) defects and pesticides, and MUS defects and solvents.

Method and Material

In this study, we evaluated whether mothers of cases with birth defects had greater potential for residential exposure to solvents, metals, or pesticides from hazardous waste sites than control mothers. We obtained residence at birth and information on potential confounders from birth certificates. Data that described hazardous waste sites and other environmental sources of exposure originated from data collected routinely for assessment or regulation. No case or control mothers were interviewed.

Case and control ascertainment. The source population for this study was all singleton births that occurred during 1983-1986 among residents of 188 counties in New York State. We excluded Long Island, New York City, and 38 sparsely populated rural counties from our study. During the period of study, addresses at birth for infants who were born in New York City hospitals, but who resided outside of New York City (approximately 5% of eligible births), were not available in a computerized form; therefore, we treated them as missing in all analyses.

We selected cases from the New York State Congenital Malformations Registry, a legislatively mandated, population-based registry that collects data on birth defects diagnosed within the first 2 y of life among residents of New York State. Defects described on narrative or computerized reports received primarily from hospitals were coded on the basis of 5-digit ICD-9 codes. We selected, by defect, two sets of cases for this study: central nervous system (CNS) defects and (2) musculoskeletal (MUS) defects. Cases born between 1983 and 1984 were also included by Geschwind et a]. in a preliminary study.[7] Central nervous system defects included cases included within ICD-9 codes 740.0-742.9 (e.g., anencephaly, spina bifida, microcephalus, encephalocele). Musculoskeletal defects included cases included within ICD-9 codes 754.0-756.9 (e.g., reduction deformities, omphalocele, syndactyly, other musculoskeletal defects). We excluded cases (n = 77) for whom there was an additional diagnosis of Trisomy 13, 18, or 21. Among the CNS group, we also excluded 35 cases who weighed less than 1 000 g at birth and who had the sole diagnosis of hydrocephaly or microcephaly; these conditions are often secondary to low birth weight and not to independent defects. We also excluded from the musculoskeletal group infants diagnosed only with hip subluxation (n = 414), which is a minor, inconsistently reported deformation. With the exception of chromosome abnormalities, we did not specifically identify or exclude genetic syndromes because genetic evaluations are not consistently conducted or reported to the registry.

We selected controls (i.e., one control group for two case groups) by randomly sampling birth certificates for residents of the 18 eligible counties (excluding multiple births). Control samples were drawn separately for the 1983-1984 and 1985-1986 births. There were approximately 3 controls per case. We excluded all eligible controls who had major malformations. We selected controls for the 1983-1984 births directly from the same controls used by Geschwind et al.[7], who, in their preliminary study, also selected their controls from a random sample of births without major malformations.

Assignment of residential locations. We used the mother's address at delivery (recorded on the birth certificate) to assign residential locations without consideration of case or control status. Study staff used an addressmatching process to convert these street addresses into latitude and longitude coordinates (geocoding) that were as close to exact as possible, based primarily on 1990 census data.[8] inadequate addresses, or those for which no matching addresses were found, were considered missing and were excluded from all analyses.

The address-matching process successfully geocoded 94% of residences in 1983-1984 and 84% of residences in 1985-1986. The match rate between the first and latter 2 birth years of the study were substantially different because, in the preliminary study, staff had removed incomplete addresses (i.e., missing post office boxes and rural routes) from the list of eligible subjects in 1983-1984. Significant characteristic differences (e.g. racial distribution, population density) among subjects with matched and unmatched addresses existed, mainly because we had great difficulty locating residences in rural and recently developed outer suburban areas. There was no difference, however, in the match rate between cases and controls, male and female infants, or among different levels of the mother's education.

Hazardous waste site identification and exposure assessment. We evaluated information on all inactive hazardous waste sites within the 18-county study area to determine potential residential exposure for pregnancies that occurred from 1982-1986. These sites, which do not currently receive hazardous waste, included those on the Federal National Priority List and those designated solely under New York State's program. We deleted sites (n = 29) if they had been delisted because no hazardous wastes at the site had been confirmed. The final number of sites was 643.

The exposure assessment methodology for this study has been described in detail previously.[9] in summary, four pathways were deemed potentially important for the teratogenic effects of residential exposure during pregnancy: (1) air vapor emanating from soil and lagoons, (2) air particulates via soil and dust, (3) groundwater exposure via private wells, and (4) groundwater and soil vapor inhalation via basements. These also constituted pathways for which sufficient data were generally available, thus enabling us to summarize potential exposure. Not all contaminant groups were applicable to all pathways. For example, solvents do not exist as air particulates. We concentrated the exposure assessment solely on solvents, metals, and pesticides, and we did not attempt to characterize potential exposure to other contaminants (e.g., polycyclic aromatic hydrocarbons, polychlorinated biphenyls).