Cancer Mortality in a Chinese Population Exposed to Hexavalent Chromium in Drinking Water : Epidemiology

Secondary Logo

Journal Logo

Original Article: Toxic Metals

Cancer Mortality in a Chinese Population Exposed to Hexavalent Chromium in Drinking Water

Beaumont, James J.*; Sedman, Richard M.*; Reynolds, Stephen D.; Sherman, Claire D.*; Li, Ling-Hong*; Howd, Robert A.*; Sandy, Martha S.*; Zeise, Lauren*; Alexeeff, George V.*

Author Information
doi: 10.1097/EDE.0b013e31815cea4c



Click on the links below to access all the ArticlePlus for this article.

Please note that ArticlePlus files may launch a viewer application outside of your web browser.

There has long been sufficient evidence from human and experimental animal studies to classify hexavalent chromium (Cr+6) compounds as carcinogenic when inhaled.1–3 Until recently there has been less evidence regarding the carcinogenicity of Cr+6 when ingested, despite the fact that large numbers of people are potentially exposed via the oral route.4 For example, Cr+6 was detected in about one-third of 7000 drinking water sources surveyed by the State of California (detection limit 1 μg/L), albeit at relatively low levels (86% <10 μg/L).5 Cr+6 contamination of groundwater can be due to natural conditions or to discharges from industrial activity such as chromium ore processing, metal plating, and use of Cr+6 as an anticorrosion agent in cooling water.6,7

Two sets of lifetime animal carcinogenicity studies of Cr+6 in drinking water have been reported. In 2007, the U.S. National Toxicology Program completed 2-year studies of cancer in F344/N rats and B6C3F1 mice given Cr+6 (sodium dichromate dihydrate) in drinking water.8 There was clear evidence of carcinogenic activity in male and female rats (squamous cell neoplasms of the oral cavity) and in male and female mice (neoplasms of the small intestine). The other lifetime animal carcinogenicity study of Cr+6 in drinking water, published in 1968, reported increased incidence of neoplasms (benign and malignant combined) of the forestomach in 3 generations of NMRI mice exposed to potassium chromate, but limitations of the study included early mortality from a virus in the first generation.9,10

Three studies of cancer rates in human populations exposed to Cr+6 in drinking water have been published in the scientific literature. A study published in 2001 found no elevation in total cancer or lung cancer mortality rates in geographic (zip code) areas in California in which drinking water was contaminated by Cr+6 in cooling water discharged from natural gas compression facilities (concentrations not reported).7 A study published in 1991 found no correlation between total cancer mortality rates and concentrations of Cr+6 in drinking water (ranging up to 10 μg/L) in 453 communities in Nebraska.11 A study published in 1987 found elevated mortality rates for total cancer, stomach cancer, and lung cancer in communities with Cr+6-contaminated water in Liaoning Province, China (Fig. 1).12 Concentrations of Cr+6 in drinking water wells were relatively high, ranging up to 20 mg/L and with the water being noticeably yellow in some wells. While the study reported rates, it did not report statistical measures of association or precision, and the original population and death count data were subsequently lost.12 Because of the relatively high exposure to Cr+6 in drinking water and the study's unique data on a digestive organ cancer (stomach cancer), the purpose of the present analysis was to reconstruct the data and estimate rate ratios (RRs) and confidence intervals (CIs) for the associations reported in 1987.

Map of a portion of northeastern China showing Liaoning Province and the city of Jinzhou, where an investigation of Cr+6 in drinking water and cancer mortality was conducted by the Jinzhon Health and Anti-Epidemic Station.


We obtained information on cancer mortality, population sizes, and water contamination in 9 geographic regions near the ferrochromium factory from reports of investigations by the local health department (the Jinzhou Health and Anti-Epidemic Station) spanning 1965–1986, from data supplied by one of the Jinzhou investigators to a US consulting firm in 1995, and from official census and population growth reports. We used the information to (1) describe the chronology, hydrogeology, and level of Cr+6 contamination of drinking water in the study regions, (2) estimate population sizes and numbers of cancer deaths in the 1970–1978 mortality observation period, and (3) calculate RRs for comparison of cancer mortality rates in the Cr+6-exposed study regions (combined) to rates in the Cr+6-unexposed study regions (combined) and Liaoning Province.

Exposure Data

A detailed chronology of events related to the Cr+6 contamination of groundwater and subsequent studies by Jinzhou investigators is presented in Table 1, and all available information on Cr+6 concentrations in drinking water in the study regions is presented in Table 2. Ferrochromium production was small-scale when it began in 1959 and had become full-scale by 1965 (there are no data about production scale for the intervening years). Residents of a nearby village (Nuer River Village) (Figs. 2, 3) noticed in mid-1964 that water in some drinking-water wells had turned yellow.13,14 Jinzhou investigators took samples of the water in January 1965, and reported detectable Cr+6. While the detection limit was 0.001 mg/L, according to telephone discussion with Dr Zhang (August 6, 1995), no information about Cr+6 analytic methods was provided. All chromium found in the water may have been assumed to have been hexavalent, because other forms of chromium are generally not water soluble.

Chronology of Events Related to the Cr+6-Contamination of Groundwater Near a Ferrochromium Factory and Studies by the Jinzhou Health and Anti-Epidemic Station (JHAS) of Drinking Water Concentrations and Cancer Mortality
All Available Data Through 1978 on Cr+6 Concentrations (mg/L) in Drinking Water in Villages Included in a Study of Cancer Mortality in the Period 1970–1978
Regions (indicated by Roman numerals) that were included in a study of cancer mortality rates in 1970–1978 by the Jinzhou Health and Anti-Epidemic Station, classified by whether Cr+6 was reported to be in drinking water. Adapted from Figure 1 of the report Study of the Effect of Environmental Pollution in the Jinzhou Area on Residents’ Health: I Mortality Analysis. The study regions were, I. Tanghezi, II. North Tanghezi (possibly 4 villages, see footnote in Table 3), III. an unnamed region containing 6 villages (Dawa, Jiantun, Huashan, Chenjiagou, Wanghutai, and Wanghugou) referred to here as North Nuer River, IV. Yaotanghezi, V. Jinchangbao, VI. Nuer River Village, VII. Shilitai, VIII. Yangxing, and IX. Wenjiatun. The original figure did not delineate boundaries between regions VII–IX.
Approximate locations in 1965 of the ferrochromium alloy factory, the Cr+6-contaminated area, and the villages of Jinchangbao and Nuer River Village. Adapted from Figure 2 of the report Chromium Contamination in the City of Jinzhou. Shaded areas indicate Cr+6 ground water contamination.

Jinzhou investigators identified 3 principal sources of Cr+6: (1) wastewater containing water-soluble Cr+6 was being discharged from the factory into a ditch that flowed into a dry riverbed and percolated into groundwater prior to reaching Yangxing village (Figs. 3, 4), (2) leaking parts in the equipment were releasing Cr+6-containing liquid waste onto the ground at the factory worksite, and (3) chromium ore residue was being stored on open ground, allowing water soluble Cr+6 to percolate into groundwater during rains.13,14 While the factory wastewater and production leaks were said to be the important sources of groundwater contamination initially, ore residue was said to be the long-term source.14

Approximate locations in 1979 of the ferrochromium alloy plant, ore residue pile, Cr+6 ground water contamination plume (shaded area), and 5 villages included in a study of cancer mortality 1970–1978. Adapted from Figure 3 of the report Chromium Contamination in the City of Jinzhou.

Early in 1965, a survey was conducted of 266 wells in the villages of Jinchangbao and Nuer River Village. A greater proportion of wells were contaminated in Nuer River Village, despite its greater distance from the factory. This was presumably because of the direction of groundwater flow (Table 2 and Fig. 3).14 Samples taken later in 1965 detected Cr+6 in the wells of villages up to 5.5 km from the alloy plant, including 5 villages included in the cancer mortality study (Table 2).12–14 By 1974 the plume was reported to have affected wells in a “long and narrow” 11.25 km2 area along the old course of the Nuer River.14Figure 4 shows the locations of the alloy plant, the old course of the Nuer River, the affected villages, and the groundwater contamination plume as of 1979, a year after the end of the mortality study.14

Sampling of drinking-water wells was most extensive in 1965 when 397 wells in the mortality study regions were assayed (Table 2). Sporadic sampling occurred in other years. Jinzhou reports did not provide the total number of wells in the regions, the fractions of wells sampled, or the method of selecting wells for study.

Pollution prevention measures started in 1965. These included a temporary halt to production, improvements of the production process, repair of equipment leaks, waterproofing workshop ground surface to prevent liquid waste from seeping underground, and addition of ferrous sulfate to ore residue, wastewater, and extracted groundwater to reduce Cr+6 to Cr+3.12–14 Jinzhou investigators reported that the control measures produced a rapid decline in groundwater Cr+6 concentrations after 1967.14

Cancer Mortality Data

Jinzhou investigators used an ecological epidemiologic study design to observe cancer mortality in 9 geographic regions near the ferrochromium factory (Fig. 2) over a 9-year period, 1970–1978.15 In 1979, the investigators completed collection of death records from police stations, which were the local repositories of death records. The method of determining the decedents’ places of residence at the time of death was not described in Jinzhou reports; this may have been based on information on the death certificate, or presumed to be the same as the police station, or perhaps obtained from interviewed survivors. Causes of death were categorized and coded according to procedures developed by the Chinese government for the National Malignant Neoplasm Survey of China, 1973–1975.15,16

Jinzhou investigators calculated cancer mortality rates by dividing the number of cancer deaths in the 9 regions by estimates of person-years-at-risk in the 1970–1978 observation period (Jinzhou reports did not present details of the person-years estimation methods).15 Rates by sex, age, and time period within the 1970–1978 observation period were not reported and are not recoverable. The investigators reported both crude and age-adjusted rates for total cancer, but reported only crude rates for stomach and lung cancer. While the reports did not identify the standard age distribution used for age adjustment of the all-cancer rates, the investigators said they used the methods of the Chinese National Malignant Neoplasm Survey of 1973–1975, which used China's age distribution in 1964 as the standard.15,16 Without explanation, rates were not presented for stomach cancer in 1 village (Nuer River Village) or for lung cancer in another village (Shilitai).15


Estimation of Age-adjustment, Populations, and Numbers of Deaths

Jinzhou reports did not provide population sizes and numbers of deaths in the study regions, and communications in 1995 from Dr. Zhang, now deceased, indicated that the original data were lost, according to telephone discussion with Dr Zhang (August 4, 1995). For the present paper we estimated the person-years-at-risk in the regions in the 1970–1978 mortality observation period by using 1982 census counts for the regions (supplied in 1995 by Dr. Zhang) and annual growth figures for the province for 1970–1982 from the National Bureau of Statistics of China.17,18 After estimating populations in each year of the observation period, we summed over the years to estimate total person-years at risk (Table 3).

Estimated Study Region Populations and Total Person-Years in the Mortality Observation Period (1970–1978), Based on 1982 Census Counts for the Study Regions, and Population Growth Estimates for Liaoning Province 1970–1982

Because age-adjusted rates were provided in Jinzhou reports only for all cancers combined, we estimated age-adjustment for stomach and lung cancer based on the reported effect of adjustment on the all-cancer rates. Specifically, for each study region we multiplied the unadjusted rates for stomach and lung cancers by the ratio of the adjusted and unadjusted rates for all cancer (Table 4).15,16 This method of estimating age-adjustment assumed that the mortality rates for all cancer, stomach cancer, and lung cancer increased similarly with age, as they did in all of China in 1973–1975 (Appendix Fig. 1, available with the online version of this article).16

Summary of Mortality Data for Study Regions Near a Ferrochromium Factory for Which Cancer Rates in the Period 1970–1978 Were Calculated by the Jinzhou Health and Anti-epidemic Station (JHAS)

We estimated the numbers of deaths in the study regions for all cancer, stomach cancer, lung cancer, and other cancers by multiplying the estimated person years by the estimated age-adjusted rates and rounding to the nearest whole number. The estimated numbers of cancer deaths are shown in Table 4.

Study Region Exposure Classification

We divided the 9 study regions into 2 groups—exposed (5 regions) and unexposed (4 regions) (Fig. 2)—based on whether or not Cr+6 was reported in drinking water. The Jinzhou investigators apparently sampled water only in the 5 regions in which Cr+6 had been found, possibly because those regions were down-gradient from the alloy plant and contamination would have been unlikely in the up-gradient study regions. We considered classifying the study regions into dose categories based on the concentration data, or on a surrogate of dose such as distance. We concluded, however, that the concentration data are too incomplete to assign dose levels, and that distance from the factory is not a reliable surrogate of dose (see Discussion below for more on exposure considerations).

RR Calculation

We compared the rates of cancer mortality in the combined regions with Cr+6 in drinking water (regions V–IX in Fig. 2) to the rates in the combined unexposed study regions (regions I–IV in Fig. 2) and to the rates in Liaoning Province as a whole by calculating RRs.

For the comparison of the rates in the exposed study regions (combined) to the rates in the unexposed study regions (combined), we assumed a binomial statistical distribution for both data sets in calculating exact mid-P 95% CIs and 2-sided hypothesis test probabilities with the PEPI Compare2 program, using the option “Rates with person-time denominators.”19

For the comparison of the rates in the exposed study regions to the rates in Liaoning Province, we obtained rates for the province in 1973–1975 (the middle of the 1970–1978 observation period), adjusted to the age distribution of the population in China in 1964, from the Atlas of Cancer Mortality in the People's Republic of China.16 The population of the province at the midpoint of the 1970–1978 observation period was about 33 million.17 We calculated a simple average of the sex-specific rates in the province to compare with the study region rates, which were available only for the sexes combined.15 We assumed a Poisson statistical distribution in calculating exact mid-P 95% CIs and 2-sided hypothesis test probabilities for 70 or fewer deaths, and approximate Fisher CIs and probabilities for more than 70 deaths, using the PEPI Describe computer program with the option “Compute SMR or indirectly standardized rate.”19


The rate of mortality from cancer of all types in the 5 study regions with Cr+6 in drinking water was only slightly elevated in comparison with the rate in the 4 nonexposed study regions (RR = 1.13; 95% CI = 0.86–1.46) and compared with the rate in the whole province (1.23; 0.97–1.53) (Table 5).

Comparison of the Cancer Mortality Rates in the Study Regions With Cr+6 in Drinking Water to the Rates in the Study Regions Without Cr+6 in Drinking Water and to the Rates in Liaoning Province

The rate of mortality from stomach cancer in the 4 exposed regions with complete stomach cancer rate data was elevated in comparison with the rate in the unexposed regions (1.82; 1.11–2.91) and in comparison with the rate in the whole province (1.69; 1.12–2.44).

Lung cancer mortality was negligibly elevated in comparison with the nonexposed regions (1.15; 0.62–2.07), but more substantially elevated in comparison with the whole province (1.78; 1.03–2.87).

Mortality from other cancers (cancers at sites other than the stomach and lung) was not elevated in the 3 exposed regions with complete rate data compared with either the nonexposed study regions (RR = 0.86) or the whole province (0.92).

The rate of stomach cancer mortality in one of the villages with Cr+6-contamined groundwater (Nuer River Village) was missing from the 1980 report on mortality and is not included in our main analyses.15 In a fax communication in 1995, Dr. Zhang estimated a rate of 28 per 100,000 per year for Nuer River Village, but he expressed concern about his estimate's accuracy.20 Neither the 1980 report on mortality nor Dr. Zhang's 1995 communication explained why the rate was missing. We repeated our stomach cancer RR calculations after including Dr. Zhang's estimate of the missing rate to determine the sensitivity of our results to this estimate. With the estimate included, the stomach cancer RR for the exposed study regions was 1.73 (1.10–2.68) compared with the unexposed regions, slightly lower than the ratio of 1.82 (1.11–2.91) without the estimate (Table 5). Similarly, with the estimate included, the RR was 1.60 (1.12–2.23) compared with the whole province, which was slightly lower than 1.69 (1.12–2.44) without the estimate (Table 5). Thus, the association of Cr+6 and stomach cancer remained when Dr. Zhang's estimate of the rate for Nuer River Village was included.


The results of our analysis are consistent with those of Zhang and Li, who reported in 1987 that rates of mortality from all cancer, lung cancer, and stomach cancer were higher in Jinzhou-area regions with Cr+6-contaminated drinking water than in the general population of the district.12 We found that mortality from “other cancers” (ie, other than stomach and lung) was not increased in regions with Cr+6-contaminated water. This provides some evidence that there was no bias that universally affected all categories of cancer (as would happen, for example, with an error in population estimation).

Our finding of elevated mortality from all cancers combined is also consistent with an analysis of the data published in 1997 that reported an increased rate for all cancers in the Cr+6 exposed study villages in comparison with Liaoning Province.21 However, that paper has since been retracted by the journal in which it appeared.22 The 1997 paper did not compare rates for stomach and lung cancer with those in the province or the unexposed study regions. It did report that within the 5 exposed villages, the village-specific rates for all cancer, stomach cancer, and lung cancer were not associated with distance to the alloy plant. On this basis, the authors dismissed a causal association of Cr+6 with cancer. The merits of using distance as a surrogate for dose are discussed below.

Stomach Cancer

The association between stomach cancer mortality and Cr+6 in drinking water in the Jinzhou data is somewhat difficult to interpret as causal, because the mortality observation period (1970–1978) ended 14.5 years after the first evidence of exposure (yellow color of drinking water reported by village residents in mid-1964) and environmental carcinogens often have a latent period of more than 15 years. However, the association could have been causal for several reasons. First, Cr+6-contamination of drinking water may have started prior to 1964 at concentrations that were not noticed by residents (groundwater Cr+6-contamination from the factory waste began in 1960).12,14 Second, the length of the latent period for environmentally caused cancers varies by agent and target organ (eg, as little as 5 years for benzene-induced leukemia) and could be relatively short for Cr+6 and stomach cancer.23 Unfortunately, there are no useful data from occupational studies on stomach cancer latency, because, at this time, there are no definitive associations of occupational exposure and stomach cancer.24 Third, Cr+6 might have interacted with exposures to other carcinogens, resulting in observation of stomach cancer within the study period.25–27

The presence of other stomach carcinogens in the general population of Liaoning Province was suggested by a relatively high rate of 20.9 stomach cancer deaths per 100,000 per year in the province in 1973–1975, which was higher than the Chinese national rate of 15.5 and the US national rate of 8.0 in the same period.16,28 Potential risk factors for stomach cancer in China include Helicobacter pylori (H. pylori) infection of the stomach, consumption of salted foods, family history of stomach cancer, and low intake of green vegetables, fresh vegetables, vitamin C, and calcium.29,30

H. pylori bacterial infection of the stomach, with a prevalence of approximately 60% to 80% historically in the rural population of China,30–32 may have been a factor in the present study for several reasons. First, H. pylori infection is a strong carcinogen for the stomach and may have initiated cancer prior to Cr+6 acting as a cocarcinogen or at a later stage in the carcinogenic process.30,33,34 Mechanisms of H. pylori carcinogenesis have been reported to include bacteria-induced genomic mutations in gastric epithelial cells (due to reduced DNA mismatch repair) and inflammation-associated epigenetic alterations (due to oxidative damage).35 Second, because H. pylori infection can cause stomach cancer, confounding bias may have occurred if the prevalence of infection differed between the Cr+6-exposed study regions and the comparison populations. Third, H. pylori produces ammonia to neutralize stomach acid and can cause lower than normal acid output as a result of injury to acid-secreting gastric glands (atrophic gastritis).36–39 At a more neutral pH, less Cr+6 would be expected to be converted to Cr+3, thus increasing the toxic potential of the Cr+6 exposure.

Lung Cancer

The lung cancer mortality rate in the combined study regions with Cr+6-contaminated groundwater was not appreciably elevated in comparison with the rate in the uncontaminated study regions (RR = 1.15; 95% CI = 0.62–2.07), but was elevated in comparison with the rate in Liaoning Province (1.78; 1.03–2.87). One reason for the differing RRs was that lung cancer mortality was elevated in a study region (Tanghezi village) that was classified as not exposed to contaminated drinking water. The estimated age-adjusted rate of lung cancer in Tanghezi of 18.3 per 100,000 persons per year was higher than the rate of 9.5 in Liaoning Province (1.96; 1.31–2.81). While this village may not have had contaminated water, it was adjacent to the alloy plant and may have experienced more respiratory exposures to Cr+6 (from ambient air and occupationally).

Potential explanations for the elevated lung cancer risk in the regions with contaminated drinking water in comparison with the province include Cr+6 in ambient air and occupational exposure to Cr+6. Jinzhou investigators reported that a large amount of airborne Cr+6 from the factory was disseminated by wind.14 The prevailing wind direction was to the north-northeast, and the villages with contaminated groundwater were located to the east of the factory, but they may have been exposed at times to airborne Cr+6.15 Also, some residents of villages with contaminated groundwater may have been employed at the factory and occupationally exposed to airborne Cr+6 (see discussion of study limitations).

Exposure Considerations

The yellowish color of the drinking water reported by residents in 1964 was consistent with the likely chemical wastes from the factory's aluminothermic production process.12,40 The production process at that time generally involved soda-ash roasting of chromite ore to produce sodium chromate that was leached from the roasted ore with water.41,42 Thus chromate ions (CrO4−2), which are yellow in aqueous solution, were likely released from the ore residue pile during rains, from factory wastewater, and from production process leaks that were said to be numerous until pollution control measures were instituted.13

The Cr+6 concentration data implied a high groundwater velocity, with Cr+6 detected 5.5 km from the source after only 6 years from initial ferrochromium production. The velocity was possible, given the hydrogeologic characteristics of the setting described by Jinzhou investigators. The factory was located on high ground in the recharge area of the aquifer, where the groundwater was as shallow as 1 m below the surface.14 Wastewater containing Cr+6 was discharged into a ditch that flowed into a dry riverbed where the depth to ground water could have been less than 1 m, essentially disposing directly into the aquifer. The wastewater flowed in the old riverbed to the west of the village of Yangxing before seeping into the riverbed entirely.14 The aquifer was described as “gravel mixed with clayey soil” and as “sandy strata.” Because the disposal occurred in a recharge zone (uplands), there would have been steeper gradients in the alluvial and fractured rock aquifers than in a valley or flat-land setting. In addition to the natural gradients, wells for drinking water, crop irrigation, and Cr+6 treatment subjected both aquifers to pumping-induced gradients. Given the factory's location on a hill in a recharge zone and the pumping of groundwater at down-gradient locations, it is reasonable to assume a relatively fast groundwater velocity.

Human exposure may have been self-limiting due to lack of palatability of water at higher concentrations of Cr+6. If so, exposures may have been lower than what might be inferred from distance or water quality data. The Jinzhou investigators reported that village residents complained about unpleasant taste, but we can't be sure that the unpleasant taste was due to Cr+6.13 The yellow color of the water may have made it unappetizing, as reported in a study of drinking water wells near a chromate factory in Mexico where residents did not want to consume water with Cr+6 concentrations above 0.5 mg/L because of its yellowish color.43

We considered whether the study regions could be classified into multiple dose categories based on available concentration data or distance from the factory. While some concentration data were available, there was little information on the total number of wells from which sampled wells were selected, how wells were selected for sampling, and the number of people using particular wells. Regarding distance from the factory, Jinzhou investigators reported a log-linear relationship between Cr+6 concentration and the distance to the factory.12 Unfortunately, the data upon which the analysis was based were not described (eg, the dates and locations of the water samples), and the investigators stated that the geographic patterns of Cr+6 concentrations changed rapidly over time due to groundwater movement and pollution control.13 Also, the fractions of wells affected in the study regions varied. For example, only a small part of Jinchangbao, the down-gradient village closest to the factory, was affected initially due to the direction of the groundwater flow (Fig. 3). Because of these factors, we concluded that the concentration data were too incomplete to classify the exposed regions by dose, and that distance from the factory was not a reliable surrogate of dose.

Limitations of the Analysis and Data

A limitation of the present analysis is that the original data on populations and mortality counts in the study regions in the 1970–1978 study period were lost and had to be reconstructed using census, population growth rate, and mortality rate data.15 Because age-adjusted rates were available only for all cancers combined, we had to estimate age-adjustment for site-specific cancer rates based on the reported effect of age-adjustment on the rates for all cancers. The statistical methods of calculating RRs and CIs assume that actual population and mortality data are used; thus our estimates do not account for unmeasured uncertainty from estimation of lost data and age-adjustment.

Another limitation is the ecological epidemiologic study design, in which there were no exposure data on individuals. Exposure was inferred based upon living in a geographic region with any Cr+6-contaminated drinking-water wells. While ecological studies can be useful for identifying associations between cancer and water contaminants, they are viewed more often as hypothesis-generating than as hypothesis-testing because of potential exposure misclassification.44,45 In the present study, it is likely that some people residing in regions classified as exposed were not actually exposed because not all wells in exposed regions were in the contamination plume. Conversely, some persons in study regions classified as unexposed may have been exposed due to previously residing in a contaminated area. Because the study regions were not wholly exposed and unexposed, the RRs may have been attenuated (closer to 1.00) in comparison with RRs based on wholly exposed and unexposed populations.45

Data on the anatomic location and the histology of the stomach cancers would have been useful but were not available. For example, it has been suggested that cancers of the gastric cardia (approximately the top inch of the stomach where it meets the esophagus) and noncardia cancers should be considered distinct entities because of potentially different risk factors, patterns of occurrence, and histopathogenesis.34,46

Another limitation is use of cancer mortality data when incidence data would have been preferable. Incidence is the event of actual interest, and mortality rates generally underestimate incidence rates. If underestimation of incidence is the same in the exposed and comparison populations, however, the RR is unbiased but less precise. Incidence data were not available for the study regions.

An additional limitation is the lack of data on employment at the ferrochromium factory. It is likely that only a small fraction of the residents of the villages with Cr+6-contaminated water worked at the factory. Dr. Zhang reported that he abandoned collecting occupational data for his health surveys after learning that over 95% of the villagers were farmers.47 Most factory workers apparently lived on the factory site; an aerial photo of the factory complex shows a large amount of on-site worker housing (Jinzhou Ferroalloy Co. Ltd., September 13, 2006). Residents of on-site housing would have been counted as residing in Tanghezi, a study region without Cr+6 in drinking water. Occupational Cr+6 exposure is most strongly associated with lung cancer and may explain the increased rates of lung cancer in Tanghezi and in study regions with Cr+6 in drinking water in comparison to the rate to the province (Tables 4 and 5).

In conclusion, our reanalysis of the Chinese data show a substantial association between stomach cancer mortality and exposure to Cr+6-contaminated drinking water in the 1970–1978 observation period, compared with nearby uncontaminated regions and with Liaoning Province. Lung cancer mortality was also increased, but only in the comparison with Liaoning Province. Possible contributions to these increases in cancer risk by occupational and ambient air exposures to Cr+6 cannot be assessed from the limited study data. While the data and our methods of analysis have limitations, they help to quantify and reinforce the associations first reported by Zhang and Li in 1987.12


We thank Craig Steinmaus, Yi Wang, Gail Krowech, and Page Painter for their valuable comments.


1. National Toxicology Program (NTP). Chromium hexavalent compounds. In: Report on Carcinogens, Eleventh Edition. U.S. Department of Health and Human Services, Public Health Service; 2005. Available at: Accessed June 12, 2007.
2. U.S. Environmental Protection Agency (EPA). Toxicological Review of Hexavalent Chromium (CAS No. 18540-29-9) In Support of Summary Information on the Integrated Risk Information System (IRIS). Washington, D.C.: U.S. Environmental Protection Agency; 1998.
3. International Agency for Research on Cancer (IARC). Chromium and chromium compounds. In: IARC Monographs on the Evaluation of Carcinogenic Risks to Humans: Chromium, Nickel, and Welding. Lyon: International Agency for Research on Cancer; 1990:49. Available at: Accessed July 9, 2007.
4. Alexeeff GV, Satin K, Painter P, et al. Chromium carcinogenicity: California strategies. Sci Total Environ. 1989;86:159–168.
5. California Department of Health Services. Chromium-6 in Drinking Water: Regulation Update. 2007. Available at: Accessed June 12, 2007.
6. Ball JW, Izbicki JA. Occurrence of hexavalent chromium in ground water in the western Mojave Desert, California. Appl Geochem. 2004;19:1123–1135.
7. Fryzek JP, Mumma MT, McLaughlin JK, et al. Cancer mortality in relation to environmental chromium exposure. J Occup Environ Med. 2001;43:635–640.
8. National Toxicology Program (NTP). Toxicology and Carcinogenesis Studies of Sodium Dichromate Dihydrate (CAS No. 7789-12-0) in F344/N Rats and B6C3F1 Mice (Drinking Water Studies). U.S. Department of Health and Human Services, NTP TR-546, NIH Publication No. 07–5887; 2007. Available at:–F1F6–975E–7D23D1519B8CD7A5#tr500. Accessed July 18, 2007.
9. Borneff J, Engelhardt K, Griem W, et al. [Carcinogens in water and soil. XXII. Experiment with 3,4-benzopyrene and potassium chromate in mice drink]. Arch Hyg Bakteriol. 1968;152:45–53 (in German).
10. Sedman RM, Beaumont J, McDonald TA, et al. Review of the evidence regarding the carcinogenicity of hexavalent chromium in drinking water. J Environ Sci Health C Environ Carcinog Ecotoxicol Rev. 2006;24:155–182.
11. Bednar CM, Kies C. Inorganic contaminants in drinking water correlated with disease occurrence in Nebraska. J Am Water Resour Assoc. 1991;27:631–635.
12. Zhang J, Li X. Chromium pollution of soil and water in Jinzhou (Chinese language with English abstract). Zhonghua Yu Fang Yi Xue Za Zhi (Chinese Journal of Preventive Medicine). 1987;21:262–264 . English translation available with the online version of this article.
13. Jinzhou Health and Anti-epidemic Station and Jinzhou Alloy Plant; Report on the Investigation of Underground Water Contamination by Discharged Chromium Containing Waste Water; 1979. Available with the online version of this article.
14. Zhang J, Li X; Chromium Contamination in the City of Jinzhou. Jinzhou Health and Anti-epidemic Station; 1986. Available with the online version of this article.
15. Zhang J, Li X; Study of the Effect of Environmental Pollution in the Jinzhou Area on Residents’ Health: I Mortality Analysis. Jinzhou Health and Anti-epidemic Station. 1980; (approximate year of publication). Available with the online version of this article.
16. Editorial Committee for the Atlas of Cancer Mortality in the People's Republic of China. Atlas of Cancer Mortality in the People's Republic of China. Shanghai, China: China Map Press; 1979.
17. National Bureau of Statistics of China. 1985 China Population Statistical Yearbook. Beijing, China: China Statistics Press; 1985.
18. Zhang J. Facsimile 8/4/95 of study region populations in 1982 census of China; 1995. Available with the online version of this article.
19. Abramson JH. WINPEPI (PEPI-for-Windows): computer programs for epidemiologists. Epidemiol Perspect Innov. 2004;1:6.
20. Zhang J. Facsimile 9/6/95 regarding stomach cancer mortality rate in Nuer River Village; 1995. Available with the online version of this article.
21. Zhang J, Li S. Cancer mortality in a Chinese population exposed to hexavalent chromium in water. J Occup Environ Med. 1997;39:315–319.
22. Brandt-Rauf P. Editorial retraction. J Occup Environ Med. 2006;48:749.
23. Agency for Toxic Substances and Disease Registry (ATSDR). Toxicological Profile for Chromium. Atlanta, Georgia: U.S. Department of Health and Human Services, Public Health Service; 2000. Available at: Accessed July 17, 2007.
24. Raj A, Mayberry JF, Podas T. Occupation and gastric cancer. Postgrad Med J. 2003;79:252–258.
25. Armitage P, Doll R. A two-stage theory of carcinogenesis in relation to the age distribution of human cancer. Brit J Cancer. 1957;11:161–169.
26. Tennant R. What is a tumor promoter? Environ Health Perspect. 1999;107:A390–A391.
27. Davidson T, Kluz T, Burns F, et al. Exposure to chromium (VI) in the drinking water increases susceptibility to UV-induced skin tumors in hairless mice. Toxicol Appl Pharmacol. 2004;196:431–437.
28. National Center for Health Statistics. Volume II—Mortality, Part A. Vital Statistics of the United States 1975. Washington: Public Health Service; 1979.
29. Kneller RW, Guo WD, Hsing AW, et al. Risk factors for stomach cancer in sixty-five Chinese counties. Cancer Epidemiol Biomarkers Prev. 1992;1:113–118.
30. You WC, Li JY, Zhang L, et al. Etiology and prevention of gastric cancer: a population study in a high risk area of China. Chin J Dig Dis. 2005;6:149–154.
31. Forman D, Sitas F, Newell DG, et al. Geographic association of Helicobacter pylori antibody prevalence and gastric cancer mortality in rural China. Int J Cancer. 1990;46:608–611.
32. Wong BC, Lam SK, Ching CK, et al. Differential Helicobacter pylori infection rates in two contrasting gastric cancer risk regions of South China. China Gastric Cancer Study Group. J Gastroenterol Hepatol. 1999;14:120–125.
33. Helicobacter and Cancer Collaborative Group.Gastric cancer and Helicobacter pylori: a combined analysis of 12 case control studies nested within prospective cohorts. Gut. 2001;49:347–353.
34. Kamangar F, Dawsey SM, Blaser MJ, et al. Opposing risks of gastric cardia and noncardia gastric adenocarcinomas associated with Helicobacter pylori seropositivity. J Natl Cancer Inst. 2006;98:1445–1452.
35. Yao Y, Tao H, Park DI, et al. Demonstration and characterization of mutations induced by Helicobacter pylori organisms in gastric epithelial cells. Helicobacter. 2006;11:272–286.
36. Barrett KE. Gastrointestinal Physiology. New York, NY: The McGraw-Hill Companies; 2006.
37. Del Valle J. Peptic ulcer disease and related disorders. In: Kasper DL, Braunwald E, Fauci AS, et al, eds. Harrison's Internal Medicine. 16th ed. New York, NY: The McGraw-Hill Companies; 2006.
38. Chen G, Fournier RL, Varanasi S, et al. Helicobacter pylori survival in gastric mucosa by generation of a pH gradient. Biophys J. 1997;73:1081–1088.
39. Pflock M, Kennard S, Finsterer N, et al. Acid-responsive gene regulation in the human pathogen Helicobacter pylori. J Biotechnol. 2006;126:52–60.
40. Papp JF. Chromium in 1995. In: Minerals Yearbook, Vol. I, Metals & Minerals. U.S. Geological Survey, GPO Stock 024-004-02545-0; 2003. Available at: Accessed July 17, 2007.
41. U.S. Environmental Protection Agency (EPA). Chromium, Ferrochromium, and Ferrochromium-silicon. In: Identification and Description of Mineral Processing Sectors and Waste Streams. Washington D.C.: Office of Solid Waste; 1995. Available at: Accessed July 17, 2007.
42. Papp JF, Lipin RR. Chromium and chromium alloys. In: Kirk-Othmer Encyclopedia of Chemical Technology. John Wiley & Sons, Inc.; 2001. Available at: Accessed July 17, 2007.
43. Armienta-Hernandez MA, Rodriguez-Castillo R. Environmental exposure to chromium compounds in the valley of Leon, Mexico. Environ Health Perspect. 1995;103(suppl 1):47–51.
44. Ayotte JD, Baris D, Cantor KP, et al. Bladder cancer mortality and private well use in New England: an ecological study. J Epidemiol Community Health. 2006;60:168–172.
45. Brenner H, Savitz DA, Jockel KH, et al. Effects of nondifferential exposure misclassification in ecologic studies. Am J Epidemiol. 1992;135:85–95.
46. Wang LD, Zheng S, Zheng ZY, et al. Primary adenocarcinomas of lower esophagus, esophagogastric junction and gastric cardia: in special reference to China. World J Gastroenterol. 2003;9:1156–1164.
47. Ye T. Communication titled “Discussion of some topics in Dr. Zhang's papers”; 1995. Available with the online version of this article.
48. Zhang J. Facsimile 8/1/95 of data from well contamination and cross-sectional acute conditions surveys as of February, 1965. 1995. Available with the online version of this article.

    Supplemental Digital Content

    © 2008 Lippincott Williams & Wilkins, Inc.