Arizona birth certificate

THERE IS SOME EVIDENCE that maternal exposure to hazardous substances via drinking water during the gestational period of a fetus may be associated with reduced birth weight of the fetus and other adverse reproductive outcomes.[1-3] Therefore, public health officials must further evaluate and determine what conditions of exposure (e.g., dose, duration of exposure) may result in adverse reproductive outcomes.

Previous waste-disposal practices at the U.S. Air Force Plant Number 44 (1958-1976) and other operations at and near the Tucson International Airport resulted in extensive trichloroethylene (TCE) contamination of the groundwater aquifer used by the city of Tucson as its sole drinking-water supply (Fig. 1).[4] In accordance with the changes required by the U.S. Safe Drinking Water Act regulations, samples were obtained from all of the city wells in 1981 and checked for the presence of volatile organic compounds (VOCs). Chemical analysis of samples taken from the eight drinking-water wells that primarily provide drinking water for people who reside in U.S. census tracts 3701, 3702, and 3703 (total 1980 U.S. Census population of 1 7,587) indicated the wells were contaminated with TCE.[5] At the end of 1981, each of the contaminated wells was taken out of service after investigators confirmed that the contamination exceeded regulatory standards.[5]

[Figure 1 ILLUSTRATION OMITTED]

There was no public information available that investigators could use to identify how long people residing in U.S. census tracts 3701, 3702, and 3703 had received contaminated drinking water. To obtain this vital piece of information, we conducted an exposure-reconstruction study to determine when the TCE entered the eight drinking-water wells.[5,6] For the exposure-reconstruction study, we used the U.S. Geological Survey analytical solutions for two-dimensional solute transport in groundwater systems to estimate the historical movement of the groundwater contamination. The results of the groundwater evaluation effort, the location of the municipal water supply wells and distribution system, and the 1980 U.S. census-tract locations were integrated with the assistance of a Geographic Information System (GIS). A GIS is a computer system that stores and links nongraphic data and information or geographically referenced data with graphic map features, thus allowing a wide range of information processing and display operations, as well as map production, analysis, and modeling. Investigators can use this system to combine and overlay data from a variety of sources--data from existing databases, exposure models, and population information (e.g., detailed maps). This type of operation can provide an accurate estimate of where people are exposed and the degree to which they are exposed. We used the GIS to assist in the selection of the target and comparison population for this study.

Material and Method

Study population. We conducted an ecological epidemiological study and used data obtained from the environmental dose-reconstruction study[5,6] and the Arizona Birth Information Tapes.[7] The environmental dose-reconstruction study established that the 1980 U.S. census tracts 3701, 3702, and 3703 would be the target population for this study.[5,6] Individuals who resided within the target census tracts received TCE-contaminated drinking water from 1978 to 1981 (estimated concentration range of [is less than] 5 micrograms of TCE per liter of water [[micro]g/l] to 107 [micro]g/l--both spatially and temporally).[5,6] We were unable to determine precisely how much TCE-contaminated drinking water each household received because the operational history of the system was and is not known. We could only determine that the contaminated wells provided the bulk of the drinking water for U.S. census tracts 3701, 3702, and 3703.[5,6]

The comparison population, which was selected from other census tracts within the city of Tucson, closely matched the socioeconomic status characteristics of the target population.[8] In addition, the comparison population obtained their drinking water from areas of the city of Tucson Drinking Water Distribution System that were not affected by the Tucson International Airport Area NPL site contaminants. To narrow the list of census tracts that could be selected as comparison tracts, we considered only those 1980 census tracts within Tucson that had similar proportions of Hispanic individuals, females of childbearing age (i.e., 15-44 y), and median household income as in the target tracts.[8] Although a precise match was not feasible, the comparison population was selected so that it was as similar to the target population as possible.

After evaluating various combinations of census tracts, we selected 1980 U.S. census tracts 200, 1200, and 4304 as the comparison tracts. The combined tracts' populations were very similar to those of the target census tracts (percentage of Hispanic origin, percentage of females of childbearing age, and percentage of year-round housing units supplied with public water). However, the comparison-tract population was more affluent than the target-tract population--as measured by 1979 median household income (i.e., $523.00 [U.S. dollars] greater, t = 2.13 vs. [t.sub.0.975] = 1.96). Although statistics indicated that the comparison population was more affluent, the $523.00 difference in median household income may not represent a real difference in buying power.

The study population comprised all of the following single live births: (1) births identified from the 1979 to 1981 Arizona birth certificate record tapes and (2) infants born to mothers whose home street addresses listed on the birth certificates were within the target or comparison census tracts. The target population comprised 1,099 babies (1,002 full-term [gestational time [is greater than] 35 wk and [is less than] 46 wk], 31 full-term low-birth-weight [LBW] infants [Arizona birth weight code of 10 or less [is less than] 2,501 gm], 60 LBW infants, and 13 very-LBW (VLBW) infants [Arizona birth weight code of 6 or less [is less than] 1,501 gm]). The comparison population comprised 877 babies (821 full-term, 31 full-term LBW, 44 LBW, and 4 VLBW).[7]

We selected the 1979-1981 time period to take advantage of computerized vital statistics records. Although the records were initially computerized in 1978, the 1978 computer records were not complete with respect to prenatal care information.

We selected a postexposure period of analysis (i.e., 1983-1985) to determine if any observed associations found for the exposure period might have been the result of some underlying census-tract-specific risk factor(s) other than exposure to TCE-contaminated drinking water. We used the identical U.S. census tracts to select the postexposure study population as was used for the exposure period (target population of 1,223 babies [1,147 full-term, 36 full-term LBW, 70 LBW, and 12 VLBW]) and for the comparison population of 1,227 babies (1,159 full-term, 35 full-term LBW, 65 LBW, and 16 VLBW).[7] Inasmuch as all of the contaminated wells were taken out of service by the end of 1981, no babies conceived after 1981 within the target and comparison U.S. census tracts were exposed to TCE via drinking water. The postexposure period permitted examination of a similar time period as the exposure period and provided for a year buffer between the exposure period and the postexposure period. This buffer ensured that none of the babies were exposed to any site-related TCE during the postexposure period.

Statistical analysis. In this investigation, we used statistical methods similar to those presented in detail elsewhere.[9-12] The following variables were found and used from Arizona birth certificate record tapes: sex of the baby, race of the baby, mother's age, mother's education, Hispanic origin of the baby, marital status, parity, month prenatal care began, number of prenatal visits, pregnancy complications, pregnancy illness, birth-weight code, gestational time (wk), and congenital anomalies identified at birth.

Because little information was available on the tapes concerning the education of the father, we removed this factor from consideration as a variable to be analyzed. In addition, any record indicating that (1) the number of prenatal visits was unknown, (2) the gestational period was 0 wk, or (3) the birth-weight code was missing were not included in the data analysis of this study (i.e., 93 records of 4,519 [2.1%]).