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.
Chlorination by-products in drinking water come from the reaction of chlorine with organic material in the water. The latter occurs naturally or originates from municipal, agricultural, and industrial wastes. Trihalomethanes (THMs), such as chloroform, bromoform, bromodichloromethane, and chlorodibromomethane, are included among the by-products; all have been found to be associated with tumors in animal studies. 1 In addition, epidemiologic studies in adults potentially exposed to THMs have reported excesses of colon, rectum, and bladder cancer. 1 Drinking water also contains metals; among them, there is arsenic, a known carcinogen that occurs naturally in some types of soils, but also as a by-product of chemical processes. When present in high concentrations in artesian well water, it has been found to be associated with urinary cancers 2 and cancers of the lung and skin. 3 Other metals found in drinking water (cadmium, chromium, lead, and zinc) have not been studied much. When administered by inhalation, cadmium is an animal lung carcinogen, but ingestion of soluble cadmium has not been associated with an increase in tumor incidence. 4 Both trivalent and hexavalent chromium occur in drinking water and the former administered orally results in increased incidence of tumors. 4 Rats fed high doses of lead in drinking water have developed renal cancers, and the International Agency for Research on Cancer lists inorganic lead as a possible human carcinogen on the basis of rodent tests. 5 Inorganic zinc is considered protective against carcinogenicity; however, it is also considered to have cocarcinogenic activities, as it may enhance or inhibit the carcinogenesis induced by other metals, such as lead and cadmium. 6 Nitrates can also be found in drinking water when private domestic wells are used but also when the source of water for communities is groundwater. Agricultural use of nitrogen-containing fertilizers and intensive animal husbandry result in increases in groundwater nitrate levels. 7 Although not toxic itself, nitrate can be reduced to nitrite in the gastrointestinal tract and reacts with dietary amines and amides to form N-nitroso compounds in the stomach. 8In vitro, the latter compounds are highly carcinogenic. A few studies have considered a link between exposure to nitrates in the water and cancer; results are not consistent. 7
We carried out a study on the relation between exposure to THMs, metals (arsenic, cadmium, chromium, lead, and zinc), and nitrates in drinking water during the prenatal and postnatal periods and the incidence of acute lymphoblastic leukemia (ALL) in children.
Subjects and Methods
Cases between 0 and 9 years of age diagnosed between 1980 and 1993 in the Province of Québec were recruited from tertiary care centers designated by government policy to hospitalize and treat children with cancer in the province. Tracing cases from these hospitals is equivalent to population-based ascertainment. To cut costs, we excluded children living in the less populated and most distant regions. Based on population denominators these regions would include approximately 10% of the provincial population. For similar reasons, from 1991 through 1993, only cases from the metropolitan Montréal region (approximately 60% of the provincial population) were included in the study. Because cancer care is covered under the universal health plan, we believe that a negligible number of children, if any, were treated outside the province.
A case was determined to have ALL (International Classification of Diseases, 9th revision, code 204.0) on the basis of clinical and biological standard criteria determined by an oncologist in a tertiary care center. To identify cases from 1980 through 1993, we used (1) hospitalization data from the provincial government’s computerized discharge data files; (2) hospitalization censuses from the respective hospitals (we checked all medical records with a relevant discharge diagnosis); (3) lists maintained by hematology-oncology laboratories of histologic data for cases; and (4) at the largest pediatric center in the province, special medical records, distinct from the hospital medical record, which are maintained in the oncology outpatient clinic.
Population-based controls (one per case) were matched on age (within 24 months), sex, and region of residence at the calendar date of the case’s diagnosis (termed the reference date) and thus were concurrently selected. These regions are based on administrative and geographical criteria determined by the government and cover wide territories, the smallest spanning more than 8,500 km2 and the largest more than 104,000 km2. With this matching, 7.5% of strata during the prenatal period and 10.2% of strata during the postnatal period (from birth to reference date) had lived in at least one similar municipality but not necessarily at the same time. The population-based controls were chosen from family allowance files. The family allowance is a government stipend awarded to all families with children living legally in Canada. This source of data was the most complete census of children for the study years. According to the expected distribution of cases based on matching criteria, a list of ten potential controls was randomly chosen from the lists.
We classified as “noneligible” the children who were adopted, lived in foster families, whose family spoke neither French nor English, who lived out of the country, whose mother was in prolonged psychiatric treatment, or whose parents were both unavailable. We identified 510 eligible cases diagnosed from 1980 through 1993 and interviewed 491 parents (96.3%); 588 eligible controls were recruited, and 493 (83.8%) parents participated. Among controls, 74 were the second controls on the list of eligible subjects, 9 were third choices, and 1 was a fourth. The others were first choices on the list of eligible subjects. Reasons for nonparticipation were a confidential telephone number, refusal to participate, or inability to trace the family. Because two controls had been interviewed for whom cases were not available, in the end two strata without cases had to be rejected, leaving 491 cases and 491 matched controls.
Sources of Data for Exposure Measurement
A 1984 bylaw requires waterworks in the Province of Québec to provide water of a quality such that it complies with preset norms. 9 Waterworks serving 50 persons or less are not constrained by any control. Because of the costs associated with testing of organic substances, except for the larger municipalities, these tests are mostly carried out by the provincial Ministry of Environment. There are 2,347 water-distribution systems coming under the bylaw in the province. 10 Sixty per cent of the distribution systems deliver untreated water to more than half a million people, and 934 systems (40%) deliver treated water to approximately 5.5 million citizens. Of the total number of distribution systems, 1,148 are under the responsibility of municipalities; 354 belong to enterprises (for example, industries); and 296 belong to institutions (for example, schools), the latter two estimated to serve less than 1% of the population; finally, 549 are privately owned and also serve about 1% of the population.
The goal of our exposure assessment was to build a municipality-exposure matrix for total and specific THMs, metals, and nitrates. To do so, we used (1) information from our interviews of parents of study subjects, (2) historical data provided by municipalities and the Ministry of Environment, and (3) a tapwater survey carried out in 1995 and 1996 in 227 homes.
Using a telephone interview, we collected information on the child’s residential history covering the pregnancy period up to the reference date. In addition, parents were asked, for each residence, what the sources of drinking water were (urban reservoir, domestic well, bottled water, water from a natural source, other, or a combination). Mothers of 98.8% of cases and 97.4% of controls answered this part of the questionnaire. Forty-eight per cent of parents were interviewed within 5 years of the date of reference (47.4% for controls and 48.4% for cases), whereas 84% and 86% of parents of cases and controls, respectively, were interviewed within 10 years of the date of reference.
A. Questionnaire to Municipalities
A list was drawn of municipalities where cases and controls had lived (from birth to the reference date), and a questionnaire was sent to all of these municipalities (N = 305). All available values for measured parameters (THMs, metals, and nitrates) were requested for the period from 1970 up to the latest date available. A total of 202 municipalities answered the questionnaire (66.2%), but only 112 sent usable data. There were data for THMs from 1981 in one large municipality and from the mid-1980s or later for the others; the larger municipalities had data on metals and nitrates from 1970 or 1976, whereas for the others the data were from the mid-1980s. Parameters were measured at point of entry in the distribution system for all municipalities; there were also measures taken at different points in the distribution system for two large municipalities. The number of annual measurements provided by the municipalities varied between 1 and 280; all measurements were dated.
B. Ministry of Environment Data
The Ministry provided data on all measurements from municipal distribution systems taken from 1986. Measurements were carried out throughout the province (except from privately owned systems and those owned by industries or institutions) at frequencies varying from once or twice a year to twice a month. A standardized protocol of collection and analysis was followed, 11 which requires water to be flushed 5 minutes before collection. The measures were taken at approximately 1.5 km of the plant in public sites such as shopping centers, schools, etc. For each water-distribution system, the status of chlorination was determined as it changed. All measurements were dated.
As part of another but yet incomplete case-control study of childhood cancer started in 1995, 227 homes of study subjects (cases and controls) were visited in 1995 and 1996 to obtain a sample of tapwater. THMs, metals, and nitrates were analyzed. Methods of collection were standardized, and water was flushed 5 minutes before collection. Methods of analysis for THMs and nitrates were based on U.S. Environmental Protection Agency methods. 12,13 Chemical analyses of metals were performed by standard methods using a Perkin-Elmer Sciex Elan-500 (Norwalk, CT) inductively coupled plasma mass spectrometer.
There were 2,100 addresses occupied during pregnancy and postnatally by 982 study subjects. The median number of addresses per study subject from the prenatal period to the reference date was 2, and the number ranged from 1 to 8. There were 67 addresses outside the province (22 cases and 21 controls had one or more such addresses); 399 municipalities were identified through residential history covering the pregnancy period to the reference date.
The matrix was developed using all of the available sources of data; it included 327 rows, one for each distinct municipality for which data from any source were available and 11 columns for the following contaminants: four specific THMs, total THMs, five metals, and nitrates. A third dimension to the matrix was the calendar year (from 1970 through 1993).
To fill the cells from the matrix, we used measurements from all three sources (municipal records, Ministry of Environment data, and tapwater survey) when available. Before 1981, there were no data available from any source for THMs. The earliest annual average measurement was imputed to that period. For other cells in the matrix, the yearly averages were used, and for cells with no data, the closest available annual average measurement (before or after a missing year) was imputed. This decision for imputation within municipalities was based on the fact that these measures were more highly correlated than across municipalities for a given year. For the prenatal period, 21% of person-years of exposure had their value assigned from the current year (that is, no imputation), whereas 52% of person-years of exposure were from the nearest 6 years or less. For the postnatal period, these figures were 26% and 61%, respectively.
Periods during which the house was using water from a domestic well were assigned a zero level for THMs, whereas in this case, levels for metals and nitrates were those of the municipality in which the residence was situated. For families reporting use of bottled water, the following strategy was used: if the household was also connected to an urban network, it was treated as if using municipal water. The rationale for this treatment is that the household was also likely to be exposed to THMs from the distribution network from showers and other uses. If the household was not connected to an urban network, all THMs were set to zero for that address; metal levels were those from the municipality, based on the rationale that whatever the source of water was for uses other than drinking, it probably was local and thus likely to have metal levels similar to that of the municipality. Gaps in residential history were handled in the following way: for average measures, the exposure was assumed to be that of the closest available average measurement, and for cumulative measures, it was assumed to be zero.
The data from tapwater samples were used in the matrix for municipalities that had no other data for specific contaminants for previous years; the closest available annual average measurements were used. There were 103 municipalities represented in our study from the tapwater survey, and the data were used for 533 addresses occupied by 111 study subjects.
Exposure Indices Used in the Analysis
A first index was as follows: over the exposure period, water never chlorinated, water chlorinated part of the time (subjects who moved during the period between cities with different chlorination status or whose chlorination status was undetermined at a particular address), and water always chlorinated. A second index was the average level of exposure over the period (sum of concentrationi × durationi in days/total duration); it was then categorized at above the 95% percentile of the distribution for cases and controls. The average level was also divided into three categories: 24th percentile or less, 25th–75th percentile, and above the 75th percentile. In addition, with the average we created a variable for the postnatal period defined as at least half the child’s life spent at 75 μg/liter or more for total THMs and likewise at 100 μg/liter. A third index was the cumulative weighted sum over the exposure period (sum of concentrationi × durationi in days), which was then categorized as above. We did not account for latency during the postnatal period.
During the prenatal period, some data were available on THMs, metals (arsenic, cadmium, chromium, and lead) as well as nitrates, and zinc for the following number of cases: 436, 463, and 331, respectively; these numbers were 422, 463, and 314 for controls in the same period. For the postnatal period, the number of cases with these data were 457, 476, and 358, respectively, whereas for controls the numbers were 441, 480, and 340. When pairs (case and control) are used in the analysis, these numbers are reduced to 378 for THMs, 435 for metals (arsenic, cadmium, chromium, and lead) as well as nitrates, and 220 for zinc for the pregnancy period and to 412, 464, and 263, respectively, for the postnatal period.
We used conditional logistic regression to estimate odds ratios and 95% confidence intervals. In addition to the matching variables, the odds ratios were adjusted for maternal age and level of schooling.
The case and control groups each included 216 girls and 275 boys. Fifteen cases were less than 1 year of age (3%); 64 cases were less than 2 years of age (13%), and 249 were less than 4 years of age (51%). In the majority of pairs (that is, 465 of 491), case and control had an age difference of less than 3 months; for other pairs, the difference ranged between 3 and 12 months or more. Other descriptive data for cases and controls are shown in Table 1; slightly more case than control mothers were in the higher age group and in the lower levels of schooling.
Table 2 shows the results for exposure to total THMs. Values at the different percentile levels (not shown) were the following for average level (μg/liter) of total THMs during pregnancy: 12.9 at the 25th percentile, 44 at the 75th, and 110 at the 95th. The corresponding values for the postnatal period were 11.8, 46.1, and 108.7, respectively. The cumulative measures were in μg/liter-days; during pregnancy, the values at the 25th, 50th, and 75th percentiles were (rounded) 3,315, 12,259, and 30,360, respectively. These values for the postnatal period were 10,056, 80,907, and 181,704, respectively. A crude index of exposure, such as water chlorinated or not, showed increased risks at both periods (pre- and postnatal) for exposure to chlorinated water part of the time but decreased risks for permanent exposure. During the prenatal period, risks associated with average level of total THMs did not show substantial increases and did not increase with level of exposure. During the postnatal period, there was a small increase associated with a cumulative exposure level above the 95th percentile. Using the 25th and 75th percentile cutoffs for the cumulative exposure index, risks did not increase with a higher level of exposure. Finally (results not shown), the adjusted odds ratio for the entire risk period (from pregnancy to diagnosis) for those who spent both periods at average level values above the 95th percentile was 1.22 (95% confidence interval = 0.53–2.81).
Table 3 gives results for the specific THMs using average level of exposure and cumulative exposure, both indices dichotomized at above the 95th percentile. Using the average level, risks were more often decreased at both periods in particular for prenatal exposure to chlorodibromomethane; however, there was a small increase in risk for postnatal exposure to bromoform. Using the cumulative exposure index, risk was increased for postnatal exposure to chloroform but substantially decreased for pre- and postnatal exposure to bromodichloromethane and for prenatal exposure to chlorodibromomethane.
Table 4 shows the results for exposure to metals and nitrates. Using the average level index, risk for the prenatal period was slightly increased for exposure to cadmium, lead, and zinc. During the postnatal period, risk was slightly increased for arsenic, cadmium, and zinc. Using the cumulative index, there were small increases in risk for arsenic and cadmium exposure postnatally as well as a substantial increase for zinc exposure. Exposure to nitrates at both periods and using average as well as cumulative indices was associated with a reduced risk.
Because there was imputation for most of the exposure data, we attempted to evaluate the impact of this strategy on our results. After carrying out analyses in which more weight was given to exposure data closer in time to the actual year of exposure, the interpretation of the results did not materially change.
Results in this study show that exposure to total or specific THMs during pregnancy does not seem to be associated with the incidence of childhood leukemia; on the contrary, exposure to some specific THMs (bromodichloromethane and chlorodibromomethane) seemed protective. Some risks were slightly increased from exposure to metals such as cadmium and zinc. During the postnatal period, risk was slightly increased for cumulative exposure to total THMs and chloroform at the highest level of the distribution (that is, above the 95th percentile) and for average exposure to bromoform at the same level. Postnatally, risks were somewhat increased with exposure to zinc, cadmium, and arsenic.
Strictly speaking and to our knowledge, there are only two studies to compare these results with and both are ecologic in nature, looking at the association between drinking water contamination and leukemia. Children were included in both studies, but separate results were not shown for them. An inverse relation between the level of THMs and the incidence of leukemia in adults was reported in the first study, 14 whereas in the second, there was little relation between ALL and exposure to THMs (we assumed that most of the ALLs were among children). 15
One of the main issues in the investigation of the effects of chlorination by-products is the measure of exposure. The challenges related to these measurements have been recently outlined. 16 These include data on specific THMs, accounting of the timing and location of the measures, and personal exposure to disinfection by-products through amount of water consumed and use of showers. Specific THM data were used in this study. We also had data on timing, as a date was attached to all measurements. These data were used in our matrix to obtain yearly values, which seems reasonable when evaluating the risk of carcinogenicity in comparison with pregnancy outcomes, for which data related to pregnancy trimesters would seem more appropriate. We also had data on location of measures: data from the municipalities were at the point of entry in the distribution system; two large municipalities and the Ministry had data taken at different points in the distribution system; and finally, we had a substantial number of samples taken from homes. We can assume that measures in this study were generally closer to the levels encountered in the homes than those taken from studies in which all measures were taken at the point of entry in the distribution system. Nevertheless, the exposure data were limited by the use of substantial imputation. We had no THM measures before 1981, although the exposure window could begin as early as 1970 for some study children; imputing a value from a later calendar period may underestimate exposure, assuming that levels of THMs went down with time as environmental controls became stricter. The same can be said of imputing a zero value for THMs to addresses with a private well, as these may contain THMs. 17 Because of missing data, in particular for the pregnancy period, a relatively substantial number of subjects were not included in the analysis. In addition, we did not have data on water consumption and bathing habits of the study children. It was reported, however, that the concentration of chloroform in the respiratory zone is substantially inferior for baths in comparison with showers, 18 which most children in this study were unlikely to take given their age.
To model exposure to disinfection by-products in a given home, we would need information on the residence time of water in the sytem, because this time affects the chlorine content. 19 The latter depends on the location of a particular house along the distribution system as well as on the flushing habits in the household before consumption of tapwater. The task of obtaining this information is complicated considerably when dealing with hundreds of distribution systems and over a long period of time, as in studies with a cancer outcome.
Whereas the transplacental carcinogenesis by organic carcinogens is well established, that of carcinogenic metals is not. Even less investigated is the postnatal effect of such exposures on infants and children. 20 Although misclassification of exposure for metals is surely present in our study, we had many data points for a large number of study subjects; without being equivalent, the measures taken in the system may be more representative of the measures in the home than those for organic volatiles.
Arsenic is the drinking water metal contaminant that has more often been associated with internal cancers 7; however, the mechanism through which this effect is likely to occur is still not well established. 21 Our results for zinc are puzzling, as this metal is not considered a carcinogen on its own but possibly as a cocarcinogen in the presence of other metals. The number of subjects with missing values for zinc was high, which could lead to a selection bias; however, the distribution of study subjects with available data was from a wide variety of municipalities in terms of size and geographical location. The levels were not especially correlated with any of the other metals in the matrix. The Montréal Urban Community (1.8 million persons) reports a substantial but unexplained increase in zinc found in the treated raw water returned to the river between 1995 and 1996. 22 Our observations may be a marker for the complex interplay of contaminants and its resulting effects. Nitrates were not associated with increased risk of leukemia, but results in previous studies of cancer have been inconsistent. 7
Despite some strengths, this study had limited ability to establish clear associations between exposure parameters and leukemia. Overall, the indications for an association between childhood leukemia and disinfection by-products as well as some metals are not strong, nor are they absent, in particular for postnatal exposure.
We thank M. Alain Riopel for his availability and his help with the Ministry data; the oncologists and hematologists Jean-Marie Leclerc (Hôpital Sainte-Justine), Mark Bernstein (Montreal Children’s Hospital), Linda Côté-Brisson (Centre Hospitalier de l’Université Laval), Josée Brossard (Centre Hospitalier de l’Université de Sherbrooke), and Reynald Simard (Centre Hospitalier de Chicoutimi) for their collaboration; and the following members of the research team: Marcelle Petitclerc, Denyse Hamer, Annie Chartier, Kim Johnston-Main, and Jean-Pierre Farant of the Environmental Laboratory of the Department of Occupational Health at McGill University.
1. Mills CJ, Bull RJ, Cantor KP, Reif J, Hrudey SE, Huston P. Workshop report: health risks of drinking water chlorination by-products: report of an expert working group. Chron Dis Can 1998; 19: 91–102.
2. Guo HR, Chiang HS, Hu H, Lipsitz SR, Monson RR. Arsenic in drinking water and incidence of urinary cancers. Epidemiology 1997; 8: 545–550.
3. Tsuda T, Babazono A, Yamamoto E, Kurumatani N, Mino Y, Ogawa T, Kishi Y, Aoyama H. Ingested arsenic and internal cancer: a historical cohort study followed for 33 years. Am J Epidemiol 1995; 141: 198–209.
4. Rojas E, Herrera LA, Poirier LA, Ostrosky-Wegman P. Are metals dietary carcinogens? Mutat Res 1999; 443: 157–181.
5. International Agency for Research on Cancer. Lead and Lead Compounds. vols. 1–42. Overall Evaluation of Carcinogenicity: An Updating of IARC Monographs. Lyon: International Agency for Research on Cancer, 1987.
6. Kasprzak KS. Effects of calcium, magnesium, zinc and iron on nickel carcinogenesis: inhibition versus enhancement. In: Hadjiliadis ND, ed. Cytotoxic, Mutagenic and Carcinogenic Potential of Heavy Metals Related to Human Environment. NATO ASI Series. 2. Environment. vol. 26. Dordrecht, the Netherlands: Kluwer Academic Publishers, 1997:93–106.
7. Cantor KP. Drinking water and cancer. Cancer Causes Control 1997; 8: 292–308.
8. Walker R. Nitrates, nitrites, and N
-nitroso compounds: a review of the occurrence in food and diet and the toxicological implications. Food Addit Contam 1990; 7: 717–768.
9. Gazette officielle du Québec. Règlement sur l’eau potable. Québec: partie 2, 116e année, no 23. 1984:2123–2129.
10. Ministère de l’Environnement et de la Faune du Québec. L’eau potable au Québec: un second bilan de sa qualité 1989–1994. Envirodoq EN970118. Québec: Ministère de l’Environnement et de la Faune du Québec, 1997.
11. Rousseau H. Suivi des concentrations de THM dans huit réseaux de distribution d’eau potable au Québec. Québec: Gouvernement du Québec, Ministère de l’environnement, Direction des écosystèmes urbains, Division des eaux de consommation, 1993.
12. U.S. Environmental Protection Agency. Measurement of purgeable organic compounds in water by capillary column gas chromatography/mass spectrometry. Method 524.2, revision 4.0. Washington DC: U.S. Environmental Protection Agency, 1992.
13. U.S. Environmental Protection Agency. The Determination of Inorganic Anions in Water by Ion Chromatography. Method 300. Washington DC: U.S. Environmental Protection Agency, 1984.
14. Fagliano J, Berry M, Bove F, Burke T. Drinking water contamination and the incidence of leukemia: an ecologic study. Am J Public Health 1990; 80: 1209–1212.
15. Foster AM, Prentice AG, Copplestone JA, Cartwright RA, Ricketts C. The distribution of leukaemia in association with domestic water quality in South West England. Eur J Can Prev 1997; 6: 11–19.
16. Swan SH, Waller K. Disinfection by-products and adverse pregnancy outcomes: what is the agent and how should it be measured? Epidemiology 1998; 9: 479–481.
17. Gibbons J, Laha S. Water purification systems: a comparative analysis based on the occurrence of disinfection by-products. Environ Pollution 1999; 106: 435–428.
18. Schlouch EL, Lévesque B, Ayotte P, Tardif R. Toxicokinetic modeling of chloroform exposure from showering and bathing in chlorinated tap water. Int J Toxicol 1999; 18: 83.
19. Rossman LA, Clark RM, Grayman WM. Modeling chlorine residuals in drinking-water distribution systems. J Environ Eng 1994; 120: 803–820.
20. Sipowicz MA, Diwan BA, Ramljak D, Buzard GS, Yu W, Waalkes MP, Rice JM, Kasprzak KS, Anderson LM. Perinatal effects of metals and cancer in offspring. In: Hadjiliadis ND, ed. Cytotoxic, Mutagenic and Carcinogenic Potential of Heavy Metals Related to Human Environment. NATO ASI Series. 2. Environment. vol. 26. Dordrecht: Kluwer Academic Publishers, 1997;123–138.
21. Abernathy CO, Liu YP, Longfellow D, Aposhian HV, Beck B, Fowler B, Goyer R, Menzer R, Rossman T, Thompson C, Waalkes M. Arsenic: health effects, mechanisms of actions, and research issues. Environ Health Perspect 1999; 107: 593–597.
22. Deschamps G. Evolution des contaminants toxiques dans les eaux usées à la station d’épuration de la Communauté Urbaine de Montréal en 1995 et 1996. Communauté Urbaine de Montréal (CUM), 1997.