the toxicology of TCE. It is hoped that the ease of identification of TCE by analytic chemical methods—that is, the ability to detect low levels of TCE—is not the basis for the association.

Dawson et al. (1990) approached the problem of associating specific health end points to chemical exposure by developing an animal model to explain the cardiac malformations associated with exposure to drinking water contamination by TCE and dichloroethylene (DCE) in Tucson, Arizona (Goldberg et al., 1990). These authors conclude that both DCE and TCE could be potent cardiac teratogens. The epidemiologic findings of cardiac malformations associated with drinking water contaminated with TCE and DCE are strengthened by the toxicologic research of Dawson et al. (1990). The value of the toxicologic confirmation of the association is relevant insofar as one of the limitations of the epidemiologic study cited by its authors was the inability to estimate individual doses because of limited sampling data, variability in exposure, lack of precise information on the geographic area of contamination, and the temporal characteristics of the contamination and exposures.

There also is ample evidence that TCE can act as a chemical neurotoxicant, as Feldman et al. (1988) have cited. However, other findings of TCE-related illnesses where the exposure levels are very low must be confirmed by additional epidemiologic findings or by toxicologic study or both.

The paper by Feldman et al. (1988) on blink reflex latency after exposure to TCE in well water from Woburn, Massachusetts, used a control group that had no stated history of occupational or environmental exposure to neurotoxicants. The authors conclude that the study subjects may have suffered subclinical cranial nerve damage as a result of their chronic ingestion of TCE contaminated water.

In studies, such as Feldman's, that rely on subjects' and controls' self-reporting of other occupational or environmental exposures, it might be useful to use hazard surveillance data on industry and occupation versus chemical exposure to ensure that neither group has an unrecognized workplace exposure that could compromise the validity of the results. The four-digit Standard Industrial Classification (SIC) code can be used to identify potential exposures that can be confirmed in interviews if necessary (Froines et al., 1986, 1989).

The ecologic study by Fagliano et al. (1990) examined the relation of the incidence of leukemia and the presence of volatile organic compounds (VOCs) in public drinking water supplies in several cities in New Jersey. The authors conclude that the results appear to suggest an association between nontrihalomethane (non-THM) VOCs and an increased incidence of leukemia among women. The incidence was el-