The authors respond: We appreciate the opportunity to discuss epidemiologic issues in our analysis of data from the study by Drs. Jiandong Zhang and Xilin Li.1,2 It is helpful that Kerger et al acknowledge that in 1995 they coauthored and helped Dr. Zhang submit a paper for publication.3 It should be clarified that this paper was eventually retracted by the journal in which it was published, and that coauthor Dr. Li is not the same Dr. Li (Xilin Li) who coauthored the original report published in a Chinese medical journal in 1987.2,4 Kerger et al say that we stated there was no gradient of exposure with distance from the alloy plant. We did not say that. We said that distance was not a reliable surrogate of dose due to rapid groundwater movement, direction of ground water flow, pollution prevention, reduction of Cr+6 in ground water with ferrous sulfate, and self-limitation of exposure due to water color and taste. Because of this, we did not attempt to correlate cancer rates for the 5 villages with contaminated water with distance from the alloy plant. Kerger et al did just such an analysis, however, reporting a regression coefficient of 0.01 (P = 0.93) for stomach cancer.5 While the coefficient was positive and technically indicative of higher stomach cancer rates in more distant villages, we think that the coefficient and its hypothesis-test probability more strongly suggest lack of association. While Kerger et al say that the main contamination zone was localized between the alloy plant and the near villages, water monitoring data indicate otherwise. Hydrogeologic investigation and water analysis found that the plume had formed a long and narrow contamination zone.6 Chromium was detected in the drinking water of the most distant down-gradient study village (Wenjiatun, 5 km from the alloy plant) in 1965, 5 years before the start of mortality observation in 1970.7 We did not report chromium concentration data from 1972 because, after 4 independent translations of the Chinese text, it was not clear which data were applicable to the cancer study.6 We did not report concentrations measured after 1978 because cancer mortality observation ended that year. Kerger et al assert we used “incorrect cancer comparison groups” in our analysis. We note precedents for the use of the same unexposed study regions and Liaoning Province as cancer comparison groups. Specifically, the unexposed study regions were defined by Jiandong Zhang and Xilin Li in their original analysis, and Kerger et al used Liaoning Province as a comparison group in their 1997 article.3,8 Kerger et al raise an important question about potential bias from differences in urbanization. In recent decades, rural stomach cancer mortality rates in China have indeed been higher than urban rates. In the 1970s, however, rural and urban rates were similar.9 During 1973-1975, the midpoint of the 1970-1978 study period, the national rural and urban stomach cancer mortality rates were 19.4 and 20.1 per 100,000 per year, respectively.10 Kerger et al say that our paper reversed the interpretation of the original researchers, when in fact our results are consistent with the findings of Jiandong Zhang and Xilin Li. These researchers reported in 1987 that mortality rates for all cancer, lung cancer, and stomach cancer in the contaminated area were increased above the rates in the general population.2 We reached very similar conclusions. James J. Beaumont Richard M. Sedman Martha S. Sandy Lauren Zeise California Environmental Protection Agency Office of Environmental Health Hazard Assessment Sacramento and Oakland, CA [email protected] Stephen D. Reynolds California Department of Conservation California Geological Survey Sacramento, CA
Read full abstract