And Intra-Ethnic Variation in Water Intake, Contact, and Source Estimates Among Tucson Residents: Implications for Exposure Analysis

www.nature.com/jea

Inter- and intra-ethnic variation in water intake, contact, and source estimates among Tucson residents: Implications for exposure analysis

BRYAN L. WILLIAMS, a,b YVETTE FLOREZ a,b AND SYDNEY PETTYGROVE a,b aDepartment of Epidemiology and Biostatistics, University of Arizona, Tuczon, Arizona bDepartment of Soil, Water, and Environmental Sciences, University of Arizona, Tuczon, Arizona

Water-related exposures among Hispanics, particularly among Mexican Americans, are relatively unknown. Exposure and risk assessment is further complicated by the absence of good time–activity data (e.g., water intake) among this population. This study attempts to provide some insight concerning water-related exposure parameters among Hispanics. Determining the extent to which non-Hispanic whites and Hispanics living in the Tucson metropolitan area differ with respect to direct water intake and source patterns is the primary purpose of this investigation. Using random digit dialing, researchers conducted a cross-sectional telephone population survey of 1183 Tucson residents. Significant ethnic variation was observed in water intake patterns among this sample, particularly in terms of source. Hispanics reported much higher rates of bottled water consumption than did non-Hispanic whites. Ethnic variation in exposure parameters such as that observed in this study increases the potential for measurement error in exposure analysis. Erroneous assumptions that exposure estimates (i.e., water intake source) are generalizable across various ethnic groups may lead to both overestimation and underestimation of contaminant exposure. Journal of Exposure Analysis and Environmental Epidemiology (2001) 11, 510–521.

Keywords: exposure, Hispanic, risk, risk assessment, water consumption, water intake activity.

Introduction Water Quality in Tucson Hispanic Community Tucson’s Hispanic population is acutely aware about water Water quality and quantity represents a crucial concern for quality issues. Over 80,000 Tucson residents, 90% of whom communities in the Sonoran Desert region. The region’s are Hispanic, live next to the Tucson International Airport water resources have along history of environmental Area (TIAA) Superfund site. The TIAA is a 16-square- problems including but not limited to insufficient amounts mile area that is positioned in the northern section of the of potable water, inadequate sewage systems, and Tucson Basin in Pima County, Arizona. Water contami- excessive levels of various contaminants (e.g., bacterio- nation in this area dates back to the early 1950s. To date at logic, nitrate, and volatile organic compounds [VOCs]) least 11 city wells and several personal wells in the region (Varady and Mack, 1995). Some of the worst water have been closed as a result of excess levels of a variety of quality problems have been observed in largely Hispanic contaminants (e.g., trichloroethylene [TCE], tetrachloro- communities near the U.S.–Mexico border region of the ethylene [PCE], dichloroethylene [1,1-DCE], chloroform, desert (Health Resources and Services Administration, benzene and chromium) (U.S. EPA, 2000). Although, the 1996; Varady and Mack, 1995). Further complicating area has been heavily remediated over the past two decades, matters is the region’s rapidly depleting groundwater the health effects of the long-term exposure to contaminated resources. Increasing demand for and decreasing supply of water are unclear. the region’s groundwater stores have inevitably led to the There is a dearth of epidemiological and exposure introduction of alternative sources of drinking water studies of Tucson’s Hispanic population. There is some (Bureau of Reclamation Phoenix, AZ, 1979; Price and evidence, however, that Tucson’s Hispanic population has Jefferson, 1997; Wilson et al., 1998). The public endured disproportionate rates of environmentally related implications if any of such introduction are obviously exposures and illness (ATSDR, 1996, 1999; Clark et al., unclear at this point in time. 1994; Goldberg et al., 1990; Kilburn and Warshaw, 1992, 1993; O’Rourke and Lebowitz, 1999). Excess morbidity among this population includes neurological impairment, 1. Address all correspondence to: Bryan L. Williams, PhD, University of Arizona, PO Box 245163, Tucson, AZ 85724. Tel.: +1-520-626-3406. systemic lupus erythematosus, multiple myeloma, several Fax: +1-520-626-8369. E-mail: [email protected] types of cancers, and congenital heart problems. Despite Received 23 August 2001. evidence of disproportionate morbidity and mortality, few Ethnic variation in water-related exposure parameters Williams et al. studies have attempted to examine potential environ- ingestion of tap water for drinking and food preparation mental exposures among this population. Additionally, no (U.S. EPA, 1997). In general, sources of water ingestion studies have examined water consumption patterns include community water supply (e.g., tap water), bottled among Tucson’s Hispanic population. In reality, many water (e.g., water vending machines), and other sources Hispanics contend that environmental inequities on (e.g., sodas, juices, coffee, etc.) (U.S. EPA, 2000). In this Tucson’s southside are frequently ignored (Clarke and study, we investigated direct water intake and total fluid Gerlak, 1998). intake only. In general, poor environmental conditions (i.e., water Several studies have assessed total fluid and tap water pollution) are frequently seen in minority communities intake among various populations (Ershow and Cantor, (Cooper, 1998; Greenberg and Schneider, 1996; Mohai 1989; Ershow et al., 1991; Roseberry and Burmaster, and Bryant, 1992; United Church of Christ, 1987; U.S. 1992; Swan and Waller, 1998; U.S. EPA, 2000; USDA, Environmental Protection Agency, 1992). Such disparities 1995). Overall, estimates of total fluid and tap water have taken their toll on the health of minority children intake and source vary substantially within and across (Metzger et al., 1995; NAEP, 1991, 1992, 1996, 1998). various studies (Burmaster, 1998a,b; Ershow and Cantor, For example, minority children suffer disproportionate and 1989; Ershow et al., 1991; Hattis et al., 1999; Roseberry excessive rates of asthma (NAEP, 1998) In fact, the and Burmaster, 1992). Table 1 provides a brief summary Natural Resources Defense Council (1997) asserts that of key water intake studies of the general population. drinking water contamination is one of the five worst EPA’s Exposure Factors Handbook provides a more health problems facing children today. Despite the comprehensive summary of water intake studies dating obvious need, relatively few studies have examined back to 1968 (U.S. EPA, 1997). exposures to contaminants in drinking water among Comparing water intake studies is difficult because these Hispanics living in the western region of the U.S. studies vary widely in methodology, sample size, popula- (Lebowitz et al., 1995, 1999a,b, 2000). However, this tions sampled, and the types of water intake estimates study represents an effort to begin a better understanding obtained (i.e., direct versus indirect intake). For example, of such exposures among Hispanics. To do so, we must some studies only examine total tap water intake (Canadian examine water intake and source patterns among the Ministry of National Health and Welfare, 1981; U.S. EPA, people living in this area. 1995) while other studies examine total fluid intake (Ershow and Cantor, 1989; Pennington, 1983; USDA, 1995; U.S. EPA, 2000). Definition of water intake and source Overall, estimates of mean water intake rates in the literature range from 1.04 to 1.63 l/person/day (Canadian Humans may ingest water either directly or indirectly. Ministry of National Health and Welfare, 1981; Cantor Direct water intake refers to direct ingestion of water as a et al., 1987; Ershow et al., 1991; Ershow and Cantor, 1989; beverage (U.S. EPA, 2000). Indirect water intake is the Gillies and Paulin, 1983; Hopkins and Ellis, 1980; Levallois ingestion of water added to food during its preparation, et al., 1998; Pennington, 1983; National Academy of excluding water that is innate to foods (U.S. EPA, 2000). Sciences, 1977; USDA, 1995; U.S. EPA, 1995, 1997, Total fluid intake refers to the ingestion of all forms of fluids 2000). As seen in Table 1, EPA’s most recent estimate of (i.e., tap water, juice, sodas) including water innate to foods average per capita consumption of water is 1.23 (l/person/ (U.S. EPA, 1997). Total tap water intake refers to the day). The 90th percentile is between 1.99 and 2.05

Table 1. Water intake studies.a

Study Total daily tap water 90th Percentile Sample size intake mean (l/person/day) (l/person/day) U.S. EPA (2000) 1.23b 2.02 15,303 Foundation for Water Research (1996) 1.14 N/A 476 USDA (1995) 1.15 N/A 15,000 Ershow and Cantor (1989) 1.41 2.28 11,731 Cantor et al. (1987) 1.30 2.40 8,000 Canadian Ministry of National Health 1.38 2.41 639 and Welfare (1981) aThe table was derived from Tables 3-31 and 3-32 in U.S. EPA Exposure Factors Handbook (U.S. EPA, 1997, pp. 3–26). bThe U.S. EPA study assessed mean total fluid intake.

Journal of Exposure Analysis and Environmental Epidemiology (2001) 11(6) 511 Williams et al. Ethnic variation in water-related exposure parameters

(l/person/day). These EPA estimates represent the most (2000) report is sampling problems and identification of recent and comprehensive data currently available. subsamples.

Direct Water Intake Studies of Hispanic Populations Source of direct water intake Very few water consumption studies delineate findings by ethnicity. Additionally, of the studies examining results by Numerous studies have also examined the various sources ethnicity, none differentiates between Mexican Americans of direct water intake (Auslander and Langlois, 1993; and Hispanics. Metzger et al. (1995) point out that International Bottled Water Association, 2000; Levallois ‘‘significant inadequacies in the collection of data on et al., 1997; The National Environmental Education and Hispanics make it difficult to make improving Hispanic Training Foundation, 1999; U.S. EPA (2000). Table 2 environmental health status a priority’’ (p. 25). In our provides a brief summary of some of the more recent water review of the literature, we could find only three studies that intake source studies: discuss water intake and source estimates among Hispanics As shown in Table 2, estimates of water intake source (Leach et al., 1999; Swan and Waller, 1998; U.S. EPA, vary substantially across studies. Some studies suggest that 1997). However, differences in the samples used in each people consume bottled water at almost the same rate as tap study preclude any meaningful comparisons. water (Levallois et al., 1997; The National Environmental First, U.S. EPA (1997) found that of 347 ‘‘Hispanic’’ Education and Training Foundation, 1999). However, the respondents, about 33% did not consume any tap water most recent EPA study does not support this suggestion daily. Approximately, 23% of Hispanic respondents re- (U.S. EPA, 2000). Additionally, comparisons across these ported drinking between 0.24 and 0.47 l/day, 21% between studies should be made cautiously, given the inconsistency 0.71 and 1.18 l/day, 12% between 1.42 and 2.13 l/day, 5% with how water source is operationally defined. Operational between 2.37 and 4.50 l/day, and 1% drank 4.73 l/day definitions of ‘‘tap water,’’ ‘‘bottled water,’’ and ‘‘other’’ (U.S. EPA, 1997). Secondly, Swan and Waller (1998) differ across studies. For example, the U.S. EPA (2000) found a sample of Hispanic women in California, on estimates define ‘‘bottled’’ water as ‘‘purchased plain average, drink about 0.50 l of tap water per day and 0.64 l of water,’’ while the Foundation for Water Research (1996) bottled water per day. The Hispanic population in the Swan study included sparkling waters in its estimates of bottled and Waller (1998) study may be somewhat comparable to water consumption. Several other studies have assessed that of the Tucson given the large proportion of Mexican various populations’ sources of water intake (American Americans living in southern California. Finally, in a study Industrial Health Council, 1994; Cantor et al., 1987; of viral hepatitis among children, Leach et al. (1999) Levallois et al., 1998; Tsang and Klepeis, 1996; Swan and examined water intake source among households across the Waller, 1998; USDA, 1995; Vena et al., 1993). However, Texas Mexico border (e.g., colonias). They found that inconsistencies in methodology, operational definitions and 42.5% of households used bottled water, 46.7% used sampling limit any meaningful comparisons across these municipal, 5.3% used spring or well water, and 5.6% did not studies. For a detailed discussion of the limitations of these provide a response. Although these studies provide some studies readers should refer to the U.S. EPA (2000) report data concerning water intake and source among Hispanics, entitled ‘‘Estimated Per Capita Water Ingestion in the these data cannot necessarily be used to predict the intake United States.’’ One of the areas discussed in the U.S. EPA activities of other Hispanic populations, particularly Mex- ican Americans.

Table 2. Water intake source studies.a Water source quality Study Tap Bottled Othera water (%) water (%) (%) Despite the differences in water intake and source estimates, U.S. EPA (2000) 75 13 10 most of these studies suggest that bottled water consump- International Bottled Water 22 13 65 tion is gradually increasing among the general populace. Association (2000) Bottled water ingestion is often thought to be protective The National Environmental 54 46 N/A (Leach et al., 1999). However, there is a growing body of Education and Training Foundation (1999) literature suggesting that the quality of bottled water is Levallois et al. (1997) 51 43 6 suspect (Allen et al., 1991; Fayad et al., 1997; Lalumandier, Auslander and Langlois 78–81 20–23 N/A 2000; McSwane et al., 1994; Natural Resources Defense (1993) Council, 1999; Page et al., 1993). Other studies raise aIncludes all beverages containing water such as sodas, juices, milk, concerns about the health effects of chlorinated tap water, coffee, tea, etc. particularly in the area of reproductive outcomes (Kallen

512 Journal of Exposure Analysis and Environmental Epidemiology (2001) 11(6) Ethnic variation in water-related exposure parameters Williams et al. and Robert, 2000; Magnus et al., 1999; Nieuwenhuijsen et variability observations for a specific real risk assessment/ al., 2000; Swan and Waller, 1998; Swan et al., 1988). It risk management problem’’ (p. 712). Other studies reveal remains unclear if either bottled or tap water is less likely to significant individual variability in water consumption (e.g., pose a risk to the public. Toxicological and exposure studies total fluid intake, tap water intake) and use (e.g., time spent are stymied by the fact that bottled water is widely showering) in the general population (Burmaster, 1998a,b; distributed, not source specific, and inconsistently regulated Ershow and Cantor, 1989; Hattis et al., 1999). Conse- (Nieuwenhuijsen et al., 2000; U.S. EPA, 2000). quently, population-specific exposure parameters should be obtained when assessing the impact of such exposures on disparate health outcomes. Water exposure activity pattern studies

Ingestion and inhalation represent the primary exposure Considerations when assessing inter-ethnic variation in pathways for contaminants in water. As mentioned exposure behaviors previously, there are a number of water ingestion studies (Ershow and Cantor, 1989; Ershow et al., 1991; Hattis et al., The role of culture and ethnicity in environmental exposure 1999; National Academy of Sciences, 1977; Roseberry and and health is poorly understood and difficult to ascertain. Burmaster, 1992; U.S. EPA, 2000; USDA, 1995). Addi- Intuitively, we might assume some inter-ethnic variation in tionally, researchers are investigating the health effects of exposure behaviors but we have little insight as to why or to tap water exposures on reproductive outcomes such as what degree such variation occurs. Our analytical tools and spontaneous abortions (Swan and Waller, 1998), hepatitis assessment approaches frequently are unable to delineate (Leach et al., 1999), premature birth and low birth weight the effects of culture on environmental exposures and health (Kallen and Robert, 2000), and birth defects (Magnus et al., outcomes. Montgomery and Carter-Pokas (1993) state, 1999). To date, there is insufficient evidence indicating tap ‘‘few multivariate analyses distinguish effects of compo- water ingestion causes excess rates of morbidity. Inadequate nents of social class (such as economic level) from the knowledge exists on the toxicity of the common mixtures of relative, joint, and independent effects of sociocultural disinfection by-products and epidemiological character- identifiers such as race or ethnicity’’ (p. 729). Anderson istics of exposures to such by-products (Nieuwenhuijsen et al. (1993) state, ‘‘multiple indicators should be used et al., 2000). when investigating the potentially confounding role of Some studies examine inhalation pathways for contam- class/socioeconomic status on variability in environmental inants in water. Shimokura et al. (1998) used a daily water exposure and health outcomes’’ (p. 681). Consequently, diary to assess exposure from showering and bathing. They identifying inter-ethnic variation in exposure behaviors found the average shower time for 79% of the women in represents only a beginning. Alternative analytical ap- their study to be 11.6 min ( ±4.0 min). While for 94% of the proaches are needed to ascertain the various determinants of men the average shower time was 10.4 min ( ±4.8 min). such behaviors (Collins et al., 1998; Williams et al., 2000). They found that 21% of women take baths with an average bath time of 22.9 min ( ±10.1 min). Only 3% of men took baths with an average bath time of 21.3 min ( ±12.4 min). Purpose and hypotheses Tap water used for showering and bathing is a source of inhalation exposure to VOCs found in the water (Jo et al., Clearly, scientists know relatively little about water-related 1990; McKone, 1989; Weisal and Jo, 1996). McKone and exposures among Hispanics, particularly Mexican Ameri- Knezovich (1991) measured the transfer efficiency of TCE cans. Exposure and risk assessment among Hispanics are from shower water into air as 61% (SD=9%). However, complicated by the absence of good time–activity data there are relatively few studies of exposures related to (e.g., water intake) on this population. The present study showering and bathing activities. attempts to provide some insight concerning water related exposure parameters among Hispanics. The purpose of this investigation is to determine the extent to which non- Lack of population specific water intake estimates Hispanic whites and Hispanics living in the Tucson metropolitan area differ with respect to direct water intake Uncertainty and individual variability are inherent compo- and source patterns. The primary null hypotheses is as nents of all exposure parameters (Hattis et al., 1999). Hattis follows. et al. (1999) point out, ‘‘variability has dimensions — of (a)Ho: Non-Hispanic whites and Hispanics living in the time, geographic area, genders, ages, or other population Tucson metropolitan area will not differ significantly with subgroups — and these dimensions often can have respect to direct water intake and source patterns. Or, Ho: important implications for the use of any particular set of Reported estimates of average and 90th percentile total daily

Journal of Exposure Analysis and Environmental Epidemiology (2001) 11(6) 513 Williams et al. Ethnic variation in water-related exposure parameters water intake and source of water intake will be equivalent Instrumentation Respondents were surveyed using the between non-Hispanic whites and Hispanics living in the Water Perception and Use Inventory (WPUI). Designed Tucson metropolitan area. specifically for telephone interviewing, the WPUI is a (b)Ho: Non-Hispanic whites and Hispanics living in the closed and open-ended item questionnaire. The WPUI Tucson metropolitan area will not differ significantly with contains five intact Likert-scaled items. A total of 80 items respect to showering and bathing behaviors. were developed and systematically field tested for use in the (c) Ho: Reported 90th percentile estimates of average WPUI. The WPUI was written both in Spanish and English. total daily water intake among this sample will not vary Standard translation protocols were used for constructing significantly from the U.S. EPA’s 90th percentile estimates the Spanish version of the instrument. The entire item pool of average total daily water intake. was field tested before final inclusion in the WPUI. Field- test data were used to evaluate items with respect to standard error of measurement. Methods Twenty-nine of the 80 WPUI items were used for this particular analysis. The subject of these 29 items included This study used a cross-sectional population survey to total daily water intake, primary source and patterns of daily obtain data. The population survey was conducted in the water intake (e.g., tap, filtered tap, and bottled), secondary following manner. Using random digit dialing, researchers source of daily water intake, potential sources of water conducted a cross-sectional telephone population survey of vapor inhalation (e.g., showering, bathing, washing car, 1183 Tucson residents. The survey was completed within lawn sprinkler, recreation), testing of tap water (e.g., results 5 weeks. Data analysis and preparation of statistical of tests), and water taste and smell. Each of the 29 items was summaries was completed within 6 weeks. analyzed separately. Intact scales were not used.

Study Region Characteristics Data Collection Data was obtained using a Computer- The city of Tucson consists of an estimated 481,100 Aided Telephone Interviewing (CATI) system. Telephone- residents (Arizona Department of Economic Security, based surveys generally evoke higher response rates than do 1999). In terms of ethnicity, Hispanics account for 28.7% mail, computer, and household surveys, particularly in large of the total population, white non-Hispanics 63.7%, African population studies (Oldendick and Link, 1994). Standard Americans 3.8%, and Native Americans 3.5% (Arizona telephone directories were used to identify phone prefixes. Department of Economic Security, 1999). Overall, ethnic Using the identified prefixes, random phone numbers were minority groups account for 36.3% of the total population generated for the sample. Calls were made between the (Arizona Department of Economic Security, 1999). Over hours of 10:00 a.m. and 8:00 p.m. (Mountain Time), 80,000 Tucson residents live within the 16-square-mile Monday through Friday, and 11:00 a.m. to 3:00 p.m. on area of the TIAA Superfund site. Approximately, 90% of Saturday and Sunday. If initial contact was not made with a these residents are Hispanic. TIAA is positioned in the given number, then the number was ‘‘called back’’ at least northern section of the Tucson Basin in Pima County, five times before it was eliminated from the sample. Arizona. Once contact was made, the interviewer undertook the following procedures. First, the interviewer stated the Sample Size purpose of the call and had the respondent confirm his or Sample size was determined through power analysis. The her telephone number, zip code, and place of residence. range of variables to be investigated, the desired level of Secondly, the interviewer solicited the participation of the precision, confidence levels, the degree of sample adult person in the household having the next birthday. variability, and the estimated proportion of households Once the respondent met the delimiting criteria, he or she in each county in the region with access to a phone were was read an informed consent and subsequently asked if he all factored into the analysis. Since a survey of precisely or she would volunteer to participate. The informed consent this nature had not been done previously, the variability of described how any respondent could terminate his or her the variables being studied within the target population is participation at any point in the interview, or decline to unknown; thus, maximum variability (P=0.5) was answer any question without risk of reprisal or could assumed. A simplified power analysis formula for reschedule the interview at a more convenient time. proportions was used for calculating sample size. In terms Investigators were required by our Institutional Review of sampling, Tucson residents living within city bounda- Board to obtain verbal Informed Consent from each ries were sampled within ±3–5% margin of sampling participant. If consent was given for the interview, the error. The power analysis indicated that approximately subject was then automatically assigned a respondent ID 1000 respondents were needed to meet the above margins code by the CATI system to identify his or her responses. of sampling error. The ID code was separated from any potential identifiers of

514 Journal of Exposure Analysis and Environmental Epidemiology (2001) 11(6) Ethnic variation in water-related exposure parameters Williams et al. the respondent (e.g., name, address, telephone number, Sample Characteristics etc.). Each respondent’s answer to specific items (e.g., Using 1990 U.S. Census data for comparison, the open-ended items) was recorded directly into the computer demographic characteristics of the sample were similar to to limit the potential for data transfer error. demographic characteristics of the study region. In terms of race, 2.6% of this sample self-identified as African Research Variables Criterion variables in this study Americans compared to approximately 3.8% of the study included the following: mean total daily water intake, water region. The actual percentage of non-Hispanic whites in the intake source, and showering/bathing activities. Predictor study region was estimated at 63.7%, while 66% of the variables included selected demographic characteristics sample self-identified as such. Native Americans constitute including ethnicity, residential location, income, and educa- approximately 3.5% of the population in the study region tional attainment. and comprised slightly more than 1% of the sample. With respect to ethnicity, 28.7% of the study region are Hispanic Treatment of Data These scores were calculated based on while 23.1% of the sample self-identified as Hispanic responses to each question, excluding ‘‘DON’T KNOW,’’ (Table 3). ‘‘REFUSED TO ANSWER’’ and ‘‘MISSING’’ responses. The sample and study region differed slightly with For analytical purposes, the Hispanic population was further respect to age distribution. In the study region, 25% of dichotomized by location of residence. Those Hispanics residents (U.S. Bureau of the Census, 1990) were between who live in zip codes 85706, 85713, and 85714 were the ages of 25 and 39 compared to 28.3% in the sample. In identified as ‘‘southside Hispanics’’ while all other the study region, 24% of residents were between the ages of Hispanics were identified as ‘‘non-southside Hispanics.’’ 40 and 64 and 8% were between the ages of 65 and 74 (U.S. Southside Hispanics constituted 41.39% of the total Bureau of the Census, 1990). Of those sampled, 40.3% Hispanic population surveyed. The rationale for this were between the ages of 40 and 64 and 10.9% were geographic comparison between non-southside and south- between the ages of 65 and 74. Both the sample and study side Hispanics is that southside residents live in closer region consist of a relatively small percentage of people proximity to the preponderance of environmental hazards over the age of 74. In the study region, 5% of residents were than do their mostly west side counterparts. As evidenced in over the age of 74, compared to 6.8% of those in the sample. the respondent characteristics, non-southside Hispanics are In terms of income, the study region and the sample better educated, have higher incomes, and live near fewer residents have similar profiles. The average median annual environmental hazards than do their southside counterparts. income for the study region is $30,000. In the sample, Two sets of comparisons were undertaken: t tests were 13.3% reported an income between $25,000 and $35,000 used to compare the mean scores of non-Hispanic whites per year and 16.1% reported an income between $35,000 and all Hispanics. To ascertain whether there was any and $50,000 per year. The sample and study regions were residency affect amongst the Hispanic population, the GLM differed somewhat with respect to gender. In the study procedure was used to compare the means of non-Hispanic region, 48% of the respondents are male. In the sample, whites, southside Hispanics, and non-Southside Hispanics. 40.4% of respondents identified as male. Although males are typically underrepresented in population surveys, the male representation in this sample is comparable to other Results large population surveys.

Response Characteristics Interviewers called respondents over a 1-month period. A total of 1183 Tucson residents completed the entire Table 3. Distribution of race/ethnicity. survey. The actual refusal rate was 20.5%. The literature suggests that refusal rates increasingly pose a source of Race/ethnicity Number Percent bias for telephone surveys (Hox and de Leeuw, 1994; White/non-Hispanic 782 66.10 Schmidley, 1986; Smith, 1995). The rate of respondent All Hispanic 273 23.08 refusal to participate in this study is comparable to that of Southside Hispanic 113 (41.39) other large population surveys (Davis and Smith, 1992; Non-southside Hispanic 160 (58.61) Luevano, 1994; Smith, 1995). In this study, interviewers Black 31 2.62 were not allowed to call an unwilling respondent back. In Asian/Pacific Islander 17 1.44 accordance with the University of Arizona Institutional Native American Indian 20 1.69 Review Board directives, researchers must adhere to a Don’t know 1 0.08 respondent’s wishes to end his or her participation at any Refused to answer 24 2.03 time during the study. Other 35 2.96

Journal of Exposure Analysis and Environmental Epidemiology (2001) 11(6) 515 Williams et al. Ethnic variation in water-related exposure parameters

Socioeconomic Stress Primary data obtained from this study 20 indicate that southside Hispanics appear to be much more 10 vulnerable to socioeconomic stressors than are white non-

Hispanics in Tucson. Southside Hispanic respondents 5 reported lower educational attainment, lower incomes, and 4 3 more years as residents of Tucson. These numbers were all statistically significantly different from those reported by 2 non-southside Hispanics and non-Hispanic whites at the 1 P<0.01 level. Thirty-three percent of the southside

Hispanic respondents reported no educational attainment .5 above a GED or high school. Responses in this same .4 category were 12.5% for non-southside Hispanics and 1.6% .3 Higher Cl .2 for non-Hispanic whites. Median income for southside Lower CI Hispanics was between $15,000 and $20,000 per year, for .1 Odds Ratio non-southside Hispanics was between $25,000 and A B C D E F G H $35,000, and for non-Hispanic whites, the median income was between $35,000 and $50,000. The mean years of Figure 2. Odds ratios and confidence intervals for water intake source residence in Tucson was 24.5 for southside Hispanics, 19.3 and contact by ethnicity. (A) Uses bottled water as primary source of drinking water. (B) Cooks mostly with bottled water. (C) Mixed for non-southside Hispanics, and 18.9 for non-Hispanic powdered drink with bottled water in past month. (D) Cooked with whites. filtered tap water in past month. (E) Mixed infant formula with bottled water in past month. (F) Usually showers less than 15 min per day. (G) Usually bathes less than 15 min per day. (H) Drinks filtered Total Daily Water Intake Some differences were observed in tap or bottled water as primary source of drinking water. water consumption and exposure reduction behaviors. Figure 1 compares total water intake between Hispanic and non-Hispanic White respondents. As shown in Figure 2, significant differences between As shown in Figure 1, on average, non-Hispanic White Hispanics and non-Hispanic whites were found in 8 of the respondents reported consuming slightly more fluids than 16 exposure parameters that we examined. The eight did Hispanic respondents (1.95 to 1.83). However, the significant odds ratios (ORs) ranged in value from 0.33 to differences were not statistically significant. The mean total 7.1. Note that Hispanics were almost four times more likely daily fluid intake for non-Hispanics and Hispanics was to drink bottled water than tap or filtered water than were 1.91. The 90th percentile estimates for mean total daily fluid their non-Hispanic counterparts. Additionally, Hispanics intake for non-Hispanics and Hispanics was 3.56. As were 67% and 62% less likely than non-Hispanic whites to shown in Figure 1, these 90th percentile estimates are bathe or shower less than 15 min, respectively. considerably higher than those provided by EPA. Significant differences were not demonstrated in the remaining eight exposure parameters. Hispanic and non- Water Intake Source and Water-Related Time Activity Hispanic respondents overall did not differ significantly in Patterns Respondents were compared in 17 different areas the following categories: washing vehicle using tap water including primary and secondary water intake source and (OR=1.31; CI 0.99–1.73); mixing infant formula with water related time activity patterns. Using a lognormal scale, unfiltered tap water in past month (OR=1.26; CI 0.51– Figure 2 shows the intake and exposure areas in which 3.10); mixing infant formula with filtered tap water in past Hispanics and non-Hispanic whites differed significantly. month (OR=1.94; CI 0.86–4.38); cooking with unfiltered tap water in past month (OR=0.84; CI 0.60–1.17); mixing powdered drink with tap water in past month (OR=1.05; CI 5 0.79–1.39); mixing powdered drink with filtered tap water Non-Hispanic White 4.5 3.56 in past month (OR=0.84; CI 0.60–1.18); using sprinkler to 4 3.5 Hispanic water lawn (OR=1.15; CI 0.87–1.51); and having a family Mean 3 1.95 1.83 2.02 member who plays in sprinkler (OR=0.93; CI 0.60–1.43). Liters/Day 2.5 2 90th Percentile for Significant differences were demonstrated between the 1.5 Entire Sample two groups with respect to the taste and smell of the water. 1 0.5 EPA 90th Percentile Hispanics were about 53% (CI 0.35 to 0.63) less likely than 0 non-Hispanic whites to report that the taste of their water Ethnicity was acceptable. Additionally, Hispanics were about 53% (CI 0.33 to 0.65) less likely than non-Hispanic whites to Figure 1. Total daily water consumption by ethnicity. report that the smell of their water was acceptable. Finally,

516 Journal of Exposure Analysis and Environmental Epidemiology (2001) 11(6) Ethnic variation in water-related exposure parameters Williams et al.

Table 4. Source and location of drinking and cooking water intake. said about exposure or risk. Nonetheless, it useful to illustrate the hypothetical impact of these observed differ- Percent z ences on exposure and risk estimates. Such hypothetical Primary drinking Tap water — no filter 30.2 356 illustrations may highlight the importance of taking in to water source Tap water — filtered 29.8 352 account inter- and intrasample variability in water intake Bottled water 37.0 437 and source estimates. For example, to what extent would Other 3.0 35 average daily dose (ADD) estimates of arsenic differ among Primary cooking Tap water — no filter 65.3 770 this sample? Using the observed 90th percentile intake water source Tap water — filtered 19.7 232 estimates and local water quality data (Tucson Water Bottled water 12.5 147 Quality Report, 2000) we can estimate the ADD of arsenic. Other 2.6 31 Using an average arsenic level of 0.003 mg/l (Tucson Water Primary location of Home 66.0 776 Quality Report, 2000) and an intake level of 3.56 l/day water consumption Work 28.7 338 (sample 90th percentile), the ADD of arsenic is estimated at during the day Other location 5.3 62 6.27Â10 À 5 mg/kg day. If we use the EPA standard estimate of water intake, we get an ADD of 3.56Â10 À 5 mg/kg day. Using the observed maximum contaminant level (MCL) of respondents did not differ significantly with respect to arsenic (0.0193 mg/l) (Tucson Water Quality Report, having their water tested. Hispanics were only about 23% 2000) and both the EPA and observed 90th percentile intake less likely than non-Hispanic whites to have ever had their estimates, the ADD estimates increase to 3.97Â10 À 4 to water tested. However, Hispanics were about 43% less 2.25Â10 À 4 mg/kg day, respectively. In this scenario, the likely (CI 0.39 to 0.82) than non-Hispanic whites to have ADD estimates approach and exceed ATSDR (2001) taken action in the past year to improve the quality of their minimal risk levels (MRL) of 3Â10 À 4 mg/kg/day for drinking water (i.e., testing, filtering, or using alternative chronic oral exposure. sources). Table 4 provides a summary of source and The observed variation in drinking water sources would location of drinking and cooking water intake among the also influence ADD of arsenic. Assuming nondetectable entire sample. levels of arsenic in bottled water and 0.003 mg/l average As shown in Table 4, overall tap water alternatives (i.e., levels in Tucson tap water and mean intake levels, the ADD bottled or filtered) were used more frequently for drinking of exclusively tap water drinkers would be 3.36Â10 À 5 mg/ than for cooking. Approximately, 68% of respondents kg day. However, given the variation in arsenic levels in reported using either filtered or bottled water as their different brands of bottled waters and varied use of each primary source of drinking water. In contrast, only 32% of brand within the sample, these ADD estimates are clearly respondents reported using either filtered or bottled water as limited. Finally, we obtained primary data on the number of their primary source of water used for cooking. Finally, 66% years respondents have lived in Tucson. On average, of respondents indicated that most of their water con- respondents reported living in Tucson for 19.3 years. If sumption during the day occurs at home. Overall, 26.2% of we substitute the 19.3 for the standard 30-year exposure respondents reported using a secondary source of drinking duration the ADD for arsenic is lowered from 6.27Â10 À 5 water. Seven percent of the respondents who primarily used mg/kg day to 4.03Â10 À 5 mg/kg day. Using ethnic specific unfiltered tap water also used a secondary source of exposure duration estimates, non-Hispanic whites would drinking water. Similarly, 7.4% of the respondents who have an ADD of 3.94Â10 À 5 mg/kg day while Hispanics primarily used filtered tap water also used a secondary would have an ADD 4.42Â10 À 5 mg/kg day. source of drinking water, while 10.8% of the respondents who primarily used bottled water also used a secondary source of drinking water. Data were not obtained concerning Summary/discussion the percentage of daily intake coming from both the primary and secondary sources of water intake. The main purpose of this investigation is to determine the extent of variation between non-Hispanic whites and Hispanics living with respect to direct water intake and The effect of sample water intake and source estimates source patterns. The groups differed significantly in their on average daily dose (ADD) of potential contaminants source of drinking water and their showering and bathing activities. However, the two groups did not differ The effect of the observed differences in source and intake significantly in their average total daily water intake. estimates on actual exposures to various water-based Overall, the reported 90th percentile estimates for average contaminants is ambiguous at best. Obviously, a multitude total daily water intake exceeded EPA’s 90th percentile of other variables must be examined before anything can be estimates by 1.54 l/day. As evidenced by these findings,

Journal of Exposure Analysis and Environmental Epidemiology (2001) 11(6) 517 Williams et al. Ethnic variation in water-related exposure parameters exposure-related behaviors such as water ingestion are to contaminants may also differ across various subgroups. population specific. The potential impact of these findings Hispanics in this sample reported spending considerably to exposure assessment warrants consideration. more time bathing and showering than did their non- These findings suggest that time–activity exposure Hispanic white counterparts. Additionally, Hispanics in this behaviors such as water intake may be subject to significant sample shower considerably longer than subjects in other inter-ethnic variation. non-Hispanic whites and Hispanics studies (Shimokura et al., 1998). Showering and bathing differed in 8 of the 17 categories of water intake activities. represents a significant source of inhalation exposure to Such disparity in activity patterns undermines the use of VOCs found in the water (Jo et al., 1990; McKone, 1989; standard parameters (i.e., ingestion rates) in exposure McKone and Knezovich, 1991; Weisal and Jo, 1996). assessment. As evidenced by these findings, Hispanics Given the southside’s history of VOC (i.e., TCE; 1,1-DCE; were much more likely than non-Hispanic whites to use TCA; MEK) contamination, it is reasonable to suggest that bottled water for various purposes. Mogelonsky (1997) and behaviors related to potential inhalation exposures should NRDC (1999) studies found similar source differences be investigated more thoroughly. between the two groups while the Swan and Waller (1998) Despite the espied differences in water intake source and did not. Interestingly, Hispanics in this sample were contact estimates, significant differences were not seen in especially protective of their infants’ water intake. His- the total daily fluid intake estimates (e.g., 90th percentile panics were about seven times more likely than non- estimates). Collectively, mean total fluid intake among this Hispanic whites to mix infant formula with bottled water, sample was slightly higher than that of EPA’s estimates despite have a much lower average income. (U.S. EPA, 2000). However, 90th percentile estimates for Further study is needed to examine inter- and intra- total intake were much higher (3.56:2.02) than EPA’s most ethnic variation in how and why various groups engage in recent estimates (U.S. EPA, 2000). Consequently, the ADD measures to protect their children and themselves from of a given water-based contaminant could potentially be environmental exposures. For example, it is plausible to higher for the upper 10th percentile of this sample. In this suggest that aggressive marketing of bottled water and past case, the high levels of fluid intake may be more of a experience with water contamination has affected the water function of climate rather than ethnicity (McNall and consumption patterns of Hispanics in this sample. Such an Schlegel, 1968). Although higher 90th percentile estimates effect has been demonstrated in Hispanic populations in the are expected in hot southwestern regions such as the past (Case, 1999; Hedden, 1996; Mogelonsky, 1997; Sonoran desert, there is little empirical evidence of such NRDC, 1999; Putterman, 1993). Yet, we know little about variation (McNall and Schlegel, 1968; U.S. EPA, 2000). the specific determinants of exposure-related behaviors and Additionally, very few studies have examined the impact of their impact on the health on certain subsets of the daily physical activity on total fluid intake (U.S. EPA, population. 2000). This study suggests that both the effects of climate, Irrespective of the rationale, the findings of this study humidity, and physical activity should be investigated imply that exposures from drinking water among this further. sample are not uniform. However, the implications of these Consistent with other studies, this study provides findings are limited. Given that we know little about the further evidence that individual factors can and do affect quality of the bottled water that the Hispanics are patterns of behaviors related to environmental exposure disproportionately ingesting, we cannot make any concrete (Burmaster, 1998a,b; Ershow and Cantor, 1989; Hattis assertions concerning exposure. Additional studies are et al., 1999). As shown in this study, water intake and needed. Wide distribution, source uncertainty, and incon- source levels do not represent constants. These behaviors sistent regulation of bottled water would likely inhibit vary within and across subpopulations, which in turn may toxicological and exposure studies among this population. result in varying exposures. As shown in this study, the However, some studies have overcome these barriers (Allen use of empirical data for exposure duration and intake can et al., 1991; Fayad et al., 1997; Lalumandier, 2000; greatly affect average daily does estimates. The impact of McSwane et al., 1994; Natural Resources Defense Council, varying time activity on exposure should be investigated 1999; Page et al., 1993). In addition, there may be some more thoroughly. Further, primary data should be impetus for examining the differential impact if any of collected in various subpopulations for exposure param- bottled versus tap water on reproductive outcomes (Kallen eters such as exposure frequency, exposure duration, and and Robert, 2000; Magnus et al., 1999; Swan and Waller, even average body weight. Ethnic variation is commonly 1998). The variability in consumption patterns observed in observed in demographic (e.g., number of years of this population provides fertile ground for conducting such residence) and personal factors (i.e., body weight). For an investigation. example, Hispanics typically have a higher average body The observed differences in bathing and showering weight and body mass index than do non-Hispanic whites behaviors suggest the possibility that inhalation exposures (Mitchell et al., 1990; National Health and Examination

518 Journal of Exposure Analysis and Environmental Epidemiology (2001) 11(6) Ethnic variation in water-related exposure parameters Williams et al.

Survey (1999) Additionally, Hispanics and non-Hispanics Canadian Ministry of National Health and Welfare. Tapwater consumption in this study even differed with respect to the amount of in Canada. Document number 82-EHD-80. Public Affairs Directo- rate. Department of National Health and Welfare, Ottawa, Canada, time they lived in the area. Future studies should collect 1981. primary data and identify secondary data sources Cantor K.P., Hoover R., Hartge P., Mason T.J., Silverman D.T., et al. concerning population specific exposure parameters (i.e., Bladder cancer, drinking water source, and tapwater consumption. A migration patterns). case–control study. J Natl Cancer Inst 1987: 79 (6): 1269–1270. This study highlights the need for additional research into Case B.M. Flooding the market (bottled water marketing). Latin Trade 1999: 7: 32. specific effects that culture and ethnicity has on exposure- Clark L.C., Giuliano A., Walsh B., Guersey de Zaplan J., Alberts D.S., related behaviors. The findings of this study should be Melster J., and Mason T.J. The Santa Cruz County Community Health interpreted cautiously given the lack of comparable studies. Survey. Prepared under contract from the Arizona Department of This study should be replicated with similar populations and Health Services, 1994. its assessments methods should be altered to better measure Clarke J.N., and Gerlak A.K. Environmental racism in the Sunbelt? A cross-cultural analysis. Environ Manage 1998: 22 (6): 857–867. water intake, contact, and source behaviors. Future studies Collins J., Schulte N., and Drolet A. Differential effect of ecologic risk could benefit from using and comparing the efficacy of factors on the low birthweight components of African-American, alternative assessment techniques commonly found in the Mexican-American, and non-Latino white infants in Chicago. J Natl dietary literature (Lindquist et al., 2000; Ocke et al., 1997; Med Assoc 1998: 90: 223–229. Valentine et al., 1994; Yokoo et al., 2001). Although it was Cooper M.H. Environmental justice: plight of the poor. CQ Res 1998: 8 (23): 537. beyond the scope of this study, self-reports of dietary Davis J.A., and Smith T.W. The NORC General Social Survey: A User’s behaviors (i.e., water ingestion) should be validated Guide. Sage, Newsbury Park, 1992. through the various techniques provided in the literature. Ershow A.G., and Cantor K.P. Total water and tapwater intake in the Additionally, future water intake, source, and contact United States: population-based estimates of quantities and sources. studies should seek corroboration of exposure through National Cancer Institute Order #263-MD-810264, Bethesda, MD, 1989. collection of biomarker data and monitoring of water Ershow A.G., Brown L.M., and Cantor K.P. Intake of tapwater and total quality. This would provide some insight as to the impact water by pregnant and lactating women. Am J Public Health 1991: 81: variation in exposure behaviors actually has on actual 328–334. contaminant exposure. Fayad N.M., Sheikheldin S.Y., et al. Migration of vinyl chloride monomer (VCM) and additives into PVC bottled drinking water. J Environ Health Sci 1997: A32 (4): 1065–1083. Foundation for Water Research. Tap water consumption in England and References Wales: findings from the 1995 national survey. Report No. DWI0771, 1996. Agency for Toxic Substances and Disease Registry (ATSDR). Final Gillies M.E., and Paulin, H.V. Variability of mineral intakes from drinking report. Disease and symptom prevalence survey Tucson International water: a possible explanation for the controversy over the relationship Airport site, Tucson, Arizona. U.S. Department of Health and Human of water quality to cardiovascular disease. Int J Epidemiol 1983: 12 Services, Public Health Service, Atlanta, 1996. (1): 45–50. Agency for Toxic Substances and Disease Registry (ATSDR). Trichlor- Goldberg S.J., Lebowitz M.D., Graver E.J., and Hicks S. An association of ethylene (TCE) subregistry registrant report baseline through follow- human congenital cardiac malformations and drinking water con- up 3. U.S. Department of Health and Human Services, Public Health taminants. J Am Coll Cardiol 1990: 16: 155–164. Service, Atlanta, May 1999. Greenberg M., and Schneider D. Environmentally Devastated Neighbor- Allen H.E., Henderson M., and Hass C.N. What’s in that bottle of water? hoods: Perceptions, Realities, and Policies. Rutgers University Press, Chemtech 1991: 21 (12): 738–742. New Brunswick, NJ, 1996. American Industrial Health Council (AIHC). Exposure factors source- Hattis D., Banati P., et al. Human interindividual variability in parameters book. AIHC, Washington, DC, 1994. related to health risks. Risk Anal 1999: 19 (4): 711–726. Anderson Y., Coulberson S., and Phelps J. Overview of the EPA/NIEHS/ Health Resources and Services Administration. Evaluation of HRSA ATSDR workshop — equity in environmental health: research issues Sanitarian Assignment in Nogales, Arizona, 1996. Available on-line: and needs. Toxicol Ind Health 1993: 9 (5): 679–683. http: // www.bphc.hrsa.dhhs.gov /borderhealth /. Retrieved Nov. 15, Arizona Department of Economic Security. Division of Employee Services 2000. and Support. Population Statistics. 1999. Available on-line: http:// Hedden C.R. Water works. 2000. Am Demogr 1996. Available on-line: www.de.state.az.us/links/economic/webpage/index.html http://www.marketingtools.com/publications/ad/ 96_ad/9601_ad/ Auslander B.A., and Langlois P.H. Toronto tap water: perception of its 9601AF04.htm. quality and use of alternatives. Can J Public Health 1993: 84 (2): 99– Hopkins S.M., and Ellis J.C. Drinking water consumption in Great Britain: 102. a survey of drinking habits with special reference to tap-water-based Bureau of Reclamation Phoenix, AZ. Central Arizona Project. Information beverages. Technical Report 137, Water Research Centre, Wiltshire, Packet No. 3, 1979. Great Britain, 1980. Burmaster D.E. Lognormal distributions for total water intake and tap Hox J.J., and de Leeuw E.D. A comparison of nonresponse in mail, water intake by pregnant and lactating women in the United States. telephone, and face-to-face surveys. Qual Quant 1994: 28: 329–344. Risk Anal 1998: 18 (1): 27–32. International Bottled Water Association (IBWA). 2000 Consumer Usage Burmaster D.E. A lognormal distribution for time spent showering. Risk Survey. Available on-line: www.bottledwater.org. Retrieved Nov. 15, Anal 1998: 18 (1): 33–36. 2000.

Journal of Exposure Analysis and Environmental Epidemiology (2001) 11(6) 519 Williams et al. Ethnic variation in water-related exposure parameters

Jo W.K., Weisel C.P., and Lioy P.J. Chloroform exposure and the health McSwane D.Z., Oleckno W.A., and Eils L.M. Drinking water quality risk associated with multiple uses of chlorinated tap water. Risk Anal concerns and water vending machines. J Environ Health 1994: 56 1990: 10 (4): 581–585. (10): 7–12. Kallen B.A.J., and Robert E. Drinking water chlorination and delivery Metzger R., Delgado J.L., et al. Environmental health and Hispanic outcome — a registry-based study in Sweden. Reprod Toxicol 2000: children. Environ Health Perspect 1995: 103 (6): 25–32. 14 (4): 303–309. Mitchell B.D., Stern M.P., et al. Risk factors for cardiovascular mortality in Kilburn K.H., and Warshaw R.H. Prevalence of symptoms of systemic Mexican Americans and non-Hispanic whites. San Antonio Heart lupus erythematosus (SLE) and of fluorescent antinuclear antibodies Study 1990: 131 (3): 423–433. associated with chronic exposure to trichloroethylene and other Mogelonsky M. Water off the shelf. Am Demogr 1997: 19 (4): 26. chemicals in well water. Environ Res 1992: 57 (1): 1–9. Mohai P., and Bryant B. (Eds.). Environmental racism: reviewing the Kilburn K.H., and Warshaw R.H. Effects on neurobehavioral performance evidence. Race and the Incidence of Environmental Hazards. Westview of chronic exposure to chemically contaminated well water. Toxicol Press, Boulder, Colorado, 1992, pp. 163–176. Ind Health 1993: 9 (3): 391–404. Montgomery L.E., and Carter-Pokas O. Health status by social class and/ Lalumandier J.A. Fluoride and bacterial content of bottled water vs. tap or minority status: implications for environmental equity research. water. JAMA 2000: 283 (24): 3183. Toxicol Ind Health 1993: 9 (5): 729–773. Leach C.T., Koo F.C., et al. The epidemiology of viral hepatitis in children National Academy of Sciences (NAS). Drinking water and health, Vol. 1. in South Texas: increased prevalence of hepatitis A along the Texas– National Academy of Sciences-National Research Council, Washing- Mexico border. J Infect Dis 1999: 180 (2): 509–513. ton, DC, 1977. Lebowitz M.D., O’Rourke M.K., et al. Population- based exposure National Asthma Education Program (NAEP). Guidelines for the measurements in Arizona: a phase I field study in support of the Diagnosis and Management of Asthma. NHLBI, Bethesda, MD, National Human Exposure Assessment Survey. J Exposure Anal 1991. Environ Epidemiol 1995: 5 (3): 297–325. National Asthma Education Program (NAEP). Guidelines for the Lebowitz M.D., O’Rourke M.K., Moschandreas D.M., Rogan S.P., Reses Diagnosis and Management of Asthma. NHLBI, Bethesda, MD, J., Gordon S.M., Needham L., Pirkle J., Ashley D., and Robertson 1992. G.L. Volatile organic compounds: exposures and biomarkers in the National Asthma Education Program (NAEP). Task force report on National Human Exposure Assessment Survey in Arizona, USA. In: the cost effectiveness, quality of care, and financing of asthma care. Raw G., Aizlewood C., Warren P. (Eds.), Indoor Air ’99: Am Rev Respir Dis 1996: 154 (2, pt. 2): S81–S130. Proceedings of the 8th International Conference on Indoor Air National Asthma Education Program (NAEP). Guidelines for the Quality and Climate, 1999, Vol. 2. Edinburgh, Espoo, Finland, 1999a, Diagnosis and Management of Asthma. NHLBI, Bethesda, MD, pp. 267–271. 1998. Lebowitz M.D., O’Rourke M.K., Moschandreas D.M., Rogan S.P., Van de National Health and Examination Survey (NHANES). Prevalence of Water P.K., Gordon S.M., Needham L., and Robertson G.L. Metals: overweight and obesity among adults: United States, 1999. Available exposures and bio-markers in the National Human Exposure on-line: www.cdc.gov/nchs/nhanes.htm. Assessment Survey (AZ NHEXAS) in Arizona, USA. In: Raw G., Natural Resources Defense Council (NRDC). Our children at risk, 1997. Aizlewood C., Warren P. (Eds.), Indoor Air ’99: Proceedings of the Available on-line: www.nrdc.org/health/kids/ocar/ocarack.asp. Re- 8th International Conference on Indoor Air Quality and Climate, 1999, trieved Nov. 15, 2000. Vol. 2. Edinburgh, 1999b, pp. 306–311. Natural Resources Defense Council (NRDC). Bottled water: pure drink or Lebowitz M.D., O’Rourke M.K., Rogan S., Reses J., Van de Water P., pure hype, 1999. Available on-line: www.nrdc.org/water/drinkin/ Blackwell A., Moschandreas D.J., Gordon S., and Robertson G. Indoor nbw.asp. Retrieved Nov. 15, 2000. and outdoor PM10 and associated metals and pesticides in Arizona. Nieuwenhuijsen M., Toledano M., et al. Chlorination disinfection Inhal Toxicol 2000: 12 (Suppl. 1): 139–144. by-products in water and their association with adverse Levallois P., Grondin J., et al. Evaluation of consumer attitudes on taste reproductive outcomes: a review. Occup Environ Med 2000: 57 (2): and tap water alternatives in Quebec. International Symposium on 73–85. Off-Flavours in the Aquatic Environment, Paris, France. Ocke M.C., Bueno de Mesquita H.B., Pos M.A., Smit H.A., van Staveren Levallois P., Guevin N., Gingras S., Levesque B., Weber J.P., and Letarte W.A., and Kromhout D. The Dutch EPIC food frequency ques- R. New patterns of drinking-water consumption: results of a pilot tionnaire. Relative validity and reproducibility for nutrients. Int J study. Sci Total Environ 1998: 209 (2–3): 233–241. Epidemiol 1997: 26 (Suppl. 1): S49–S58. Lindquist C.H., Cummings T., and Goran M.I. Use of tape-recorded food Oldendick R.W., and Link M.W. The answering machine generation: who records in assessing children’s dietary intake. Obes Res 2000: 8 (1): are they and what problem do they pose for survey research? Public 2–11. Opin Q 1994: 58 (2): 264–273. Luevano P. Response rates in the national election studies. Technical O’Rourke M.K., and Lebowitz M.D. Indoor bioaerosol contaminants: Report No. 44, National Elections Studies, Ann Arbor, Michigan, Environmental Toxicants: Human Exposures and their Health Effects. 1994. New York: van Nostrand Reinholt, 1999, pp. 449–480. Magnus P., Jaakkola J., et al. Water chlorination and birth defects. Page B.D., Conacher H.B.S., et al. Survey of bottled drinking-water sold Epidemiology 1999: 10 (5): 513–517. in Canada. J AOAC Int 1993: 76 (1): 26–31. McKone T.E. Household exposure models. Toxicol Lett 1989: 49 (2–3): Pennington J.A.T. Revision of the total diet study food list and diets. JAm 321–339. Diet Assoc 1983: 82: 166–173. McKone T.E., and Knezovich J.P. The transfer of trichloroethylene Price S., and Jefferson F.T. Corrosion control strategies for changing water (TCE) from a shower to indoor air: experimental measurements supplies in Tucson, Arizona. J New Engl Water Works Assoc 1997: 111 and their implications. J Air Waste Manage Assoc 1991: 41 (6): (3): 285–293. 832–837. Putterman D. Three examples of what Hispanic marketing can do. McNall P.E., and Schlegel J.C. Practical thermal environmental limits for Brandweek 1993: 34: 26. young adult males working in hot, humid environments. ASHRAE Roseberry A.M., and Burmaster D.E. Lognormal distribution for water Trans 1968: 74: 226–235. intake by children and adults. Risk Anal 1992: 12: 99–104.

520 Journal of Exposure Analysis and Environmental Epidemiology (2001) 11(6) Ethnic variation in water-related exposure parameters Williams et al.

Schmidley A.D. How to overcome bias in telephone survey. Am Demogr U.S. EPA. The National Human Activity Pattern Survey (NHAPS): use 1986: 8: 50–55. of nationwide activity data for human exposure assessment. U.S. Shimokura G.H., Savitz D.A., and Symanski E. Assessment of water use EPA, 1995. for estimating exposure to tap water contaminants. Environ Health U.S. EPA. Exposure factors handbook. National Center for Environmental Perspect 1998: 106 (2): 55–59. Assessment, Office of Research and Development, Washington, DC. Smith T.W. Trends in non-response rates. Int J Public Opin Res 1995: 7 EPA/600/P-95/002Fabc, August 1997. (2): 157–171. U.S. EPA. Estimated per capita water ingestion in the United States. Swan S.H., and Waller K. Disinfection by-products and adverse Office of Science and Technology, Office of Water, Washington, pregnancy outcomes: what is the agent and how should it be DC, 2000. measured? Epidemiology 1998: 9 (5): 479–481. United States Department of Agriculture (USDA). Food and nutrient Swan S.H., Deane M., et al. Water exposure and pregnancy outcomes. intakes by individuals in the United States, 1 day, 1989–91. United Epidemiological Studies and Surveillance Section, California Depart- States Department of Agriculture, Agricultural Research Service. NFS ment of Health Services, Berkeley, California, 1988. Report No. 91-2, 1995. The National Environmental Education and Training Foundation. 1999 Valentine J.L., Cebrian M.E., et al. Daily selenium intake estimates for Safe Drinking Water Report Card. Available on-line: www.neetf.org/. residents of arsenic-endemic areas. Environ Res 1994: 64 (1): 1–9. Retrieved Nov. 15, 2000. Varady R.G., and Mack M.D. Transboundary water resources and public Tsang A.M., and Klepeis N.E. Results tables from a detailed analysis of health in the U.S.–Mexico border region. J Environ Health 1995: 57 the National Human Activity Pattern Survey (NHAPS) response. (8): 8–14. Draft Report prepared for the U.S. Environmental Protection Agency Vena J.E., Graham S., Freudenheim J., et al. Drinking water, fluid intake by Lockheed Martin, Contract No. 68-W6-001, Delivery Order No. and bladder cancer in western New York. Arch Environ Health 1993: 13, 1996. 48: 191–198. Tucson Water Quality Report. 2000. Available on-line: http://www.ci. Weisal C.P., and Jo W.K. Ingestion, inhalation, and dermal exposures to tucson.az.us/water/water_quality/emerging_wq_issues/arsenic/ar- chloroform and trichloroethane from tap water. Environ Health senic.htmv. Perspect 1996: 104: 48–51. United Church of Christ (UCC), Commission for Racial Justice. Toxic Williams B., Pennock-Roman M., and Suen H. Assessing the differential waste and race in the United States: a national report on the racial and impact of social and environmental exposures on the health status of socioeconomic characteristics of communities with hazardous waste Hispanic populations: a case for hierarchical linear modeling. Socio- sites. New York, NY, 1987. econ Plann Sci. (In press). U.S. Bureau of the Census. Census of population and housing for Wilson L.G., Matlock W.G., and Jacobs K.L. Hydrogeologic uncertainties the State of Arizona. U.S. Bureau of the Census, Washington, and policy implications: the Water Consumer Protection Act of DC, 1990. Tucson. Hydrogeol J 1998: 6 (1): 3–14. U.S. Environmental Protection Agency (EPA/NIH). Respiratory health Yokoo E.M., Valente J.G., Sichieri R., and Silva E.C. Validation and effects of passive smoking: lung cancer and other disorders. EPA/600/ calibration of mercury intake through self-referred fish consumption in 6-90/006F. Smoking and Tobacco Control Monograph No. 4. NCI, riverine populations in Pantanal Mato-grossense, Brazil. Environ Res Bethesda, MD, 1993. 2001: 86 (1): 88–93.

Journal of Exposure Analysis and Environmental Epidemiology (2001) 11(6) 521