CDC
CCHTFF5 rcn
 V-/EPA
United States
Environmental Protection   Office of Water   EPA 815-R-99-001
Agency           4607       January 1999


Health Effects from Exposure to

High Levels of Sulfate in

Drinking Water Study

-------
                                         IRBINFO # EHHEHSB979
                                         CDC Protocol # 1779
Health Effects from Exposure to High Levels of
        Sulfate in Drinking Water Study
                  January 21, 1999
                Health Studies Branch
       Environmental Hazards and Health Effects
       National Center for Environmental Health
       Centers for Disease Control and Prevention
       Office of Drinking Water and Ground Water
         U.S. Environmental Protection Agency
                      Page 1

-------
Executive Summary

       Sulfate is a substance that occurs naturally in drinking water.  Health concerns regarding
sulfate in drinking water have been raised because of reports that diarrhea may be associated with the
ingestion of water containing high levels of sulfate. Of particular concern are groups within the
general population that may be at greater risk from the laxative effects of sulfate when  they
experience an abrupt change from drinking water with low sulfate concentrations to drinking water
with high sulfate concentrations.

       There are very few scientific reports that address sulfate concentration in drinking water and
the effects it may have on the health of those individuals who are exposed. Furthermore, the concerns
regarding sensitive populations are based solely on case studies and anecdotal reports.  One  such
potentially sensitive population is infants receiving their first bottles containing tap water, either as
water alone or as formula mixed with water.  Other groups  of people who could potentially be
adversely affected by water with high sulfate concentrations include transient populations  (i.e.,
tourists, hunters, students, and other temporary visitors) and people moving  from areas with low
sulfate concentrations in drinking water into areas  with high concentrations.

       The objective of the present study was to provide additional information regarding whether
sensitive populations (infants and transients) may be adversely  affected by sudden exposure to
drinking water containing  high levels of sulfate. Specifically,  we designed a field investigation to
recruit 880 infants exposed to naturally occurring high levels of sulfate in the drinking water provided
by public water systems and an experimental trial of exposure  in adults.

       We planned a prospective cohort study of infants born in geographic areas with naturally
occurring high levels of sulfate in the  drinking water provided by public water systems in New
Mexico, South Dakota, and Texas.  Infants were to be enrolled at birth and followed for four weeks
to determine if there was an association between exposure to drinking water containing varying levels
of sulfate and reported cases of diarrhea.

       We conducted a pilot study of the planned recruitment methods and study instruments in four
counties in South Dakota with high levels of sulfate in the drinking water provided by the public
water systems. We approached 72 pregnant women from these  counties (served by three community
health clinics) about participating in the study, but only eight were  eligible based on the study
eligibility criteria. Of the eight eligible women, three refused  to participate.  Only one of the five
women who agreed to participate completed all of the study activities.

       Because  we experienced recruiting problems during the pilot study, we  developed a  self-
administered questionnaire (SAQ) to examine tap water use. The  questionnaires  were provided to
all women who received care during a two-week period from one of 32 Women, Infants and Children
(WIC) clinics in New Mexico, South Dakota, and Texas.  The clinics were located in geographic
areas with a range of sulfate levels (from <100 mg/L to > 1000 mg/L) in the drinking water provided
by public water systems. The SAQ asked questions about the source of the women's home tap water,

                                          Page   2

-------
what mothers of infants < 3 months old were currently feeding their babies, and how pregnant women
planned to feed their infants.

       Of the  951 pregnant women who responded to the SAQ, more than half (582 or 61%)
reported planning to breast-feed their infants and only 141 (15%) reported that they planned to use
infant formula mixed with their home tap water to feed their infants. Of the 542 mothers with infants
< 3 months of age, 272 (50%) reported having breast fed their infants.

       To determine how many of the 1388 women who completed the SAQ would have been
eligible to participate in our study based on the drinking water source and use criteria, we examined
the responses of the 1164 women who received their tap water from public water systems and who
did not have filters on their home taps.  We  found that 403 women with infants < 3 months of age
and 761 pregnant women met these eligibility criteria.  The largest numbers of women who had used
or were planning to use tap water to mix infant formula were in the 4 counties with average sulfate
levels < 250 mg/L. Of the 365 pregnant women in areas with sulfate levels > 250 mg/L, only 39
(11%) planned to use infant formula mixed with their tap water. Of the 183 women with infants <3
months  old in  areas with sulfate levels > 250 mg/L, only 35 (19%) reported having used infant
formula mixed with their tap water.  Thus, of the 548 pregnant women and women with infants < 3
months  old, only 74 infants were or would be exposed to tap water containing > 250 mg/L sulfate.
These results are consistent with our findings during the pilot study and indicate that only a very small
number of women who live in areas with high levels of sulfate in the tap water provided by public
water systems plan to give this water to their infants.

       Another population potentially sensitive to abrupt exposure to high levels  of sulfate in
drinking water is transient adults (students, visitors, hunters, etc.).  To study the effects in adults of
suddenly changing drinking water sources from one that has little or no sulfate to one that is high in
sulfate, we conducted an experimental study involving volunteers from Atlanta, Georgia, including
CDC employees and employees at the U.S. EPA Region IV office.  Volunteers  were randomly
assigned to one of five sulfate exposure groups (i.e., 0, 250, 500, 800, or 1200 mg/L sulfate from
sodium sulfate in bottled drinking water) and were provided with bottled drinking water for six days.
The bottled water for days 1, 2, and 6 contained plain water, while the bottles for days 3 through 5
contained water with added sulfate. The unfinished or empty bottles were returned and weighed to
determine how much water was consumed each day. Volunteers were blinded to the level of sulfate
in their drinking water.

       One hundred and five study participants were divided among the dose groups as follows:  24
received 0 mg/L sulfate; 10 received 250 mg/L sulfate; 10 received 500 mg/L sulfate;  33 received 800
mg/L sulfate; and 28  received 1200 mg/L sulfate.  We analyzed the number of bowel movements
recorded each  day by study participants.  There were no  statistically significant differences in the
bowel movements among the groups on days 3, 4, 5, or 6. There were also no statistically significant
differences in the bowel movements reported when comparing days 1 and 2 (the days when there was
no sulfate in the water) with days 3, 4, and 5 within each  dose group.
                                         Page

-------
       To examine the data for a trend toward increased frequency of reports of diarrhea with
increased dose of sulfate, we included the dose as an ordinal variable in a logistic regression model
of osmotic diarrhea.  There was no statistically significant increase in reports of diarrhea with
increasing dose (one-sided p = 0.099).

       The overall purpose of these studies was to examine the association between consumption
of tap water containing high levels  of sulfate and reports  of osmotic  diarrhea in susceptible
populations (infants and transients). We were unable to conduct a study of infants because we could
not identify enough exposed individuals from which to draw a study population. The results of our
SAQ examining tap water use indicated that more than half of the pregnant women who completed
the survey planned to breast-feed their infants.  Of those who planned to use formula mixed with
water, most did not plan to use tap water to mix the formula. In our experimental trials with adult
volunteers, we did not find an association between acute exposure to sodium sulfate in tap water (up
to 1200 mg/L) and reports of diarrhea.

       We were not able to conduct the dose-response studies that were requested as part of the Safe
Drinking Water Act Amendments of 1996.  Instead, we convened an expert workshop whose
members reviewed the available literature and addressed a series of questions about the health effects
from exposure to sulfate in drinking water: 1) Do reported studies suggest that a certain sulfate level
would not be likely to cause adverse  effects?, 2)  Does the literature support acclimatization or
adaptation?, 3) Can an infant dose-response study be done anywhere in the U.S. or Canada?, 4) Is
there enough scientific evidence that there adverse health effects from  sulfate in drinking water to
support regulation?  The workshop was held in Atlanta, Georgia on September 28,  1998.   The
workshop report accompanies this document.
                                         Page  4

-------
Introduction

       Sulfate is a substance that is often found in drinking water. Health concerns regarding sulfate
in drinking water have been raised because of reports of diarrhea associated with the ingestion of
water containing high levels of sulfate.  Available data suggest that people acclimate rapidly to the
presence of sulfates in their drinking water.  However, there are groups within the general population
that may be at greater risk from the laxative effects of sulfate when they experience an abrupt change
from  drinking  water  with low  sulfate  concentrations to drinking water with high sulfate
concentrations.

       One such potentially sensitive population is infants receiving their first bottles containing tap
water, either as water alone or as formula mixed with water. A series of three case histories from
Saskatchewan reported by Chien etal. (1968) suggested that infants may experience gastroenteritis,
including diarrhea and dehydration, upon their first exposure to water that contains high levels of
sulfate. The three infants  discussed in the  report were symptom-free until their families moved to
areas with water supplies that contained high levels of sulfate (650 to 1150 mg/L).  Interestingly, the
infants developed diarrhea when they were  given water from these new sources. Stools from two of
the infants tested negative for bacterial pathogens, ova, and parasites; and the diarrhea subsided when
alternative  water  sources were used.  In light of these reports, the authors suggested that sulfate
levels are important with respect to their laxative effect on babies. They recommended that water be
screened for sulfate content if a sample is submitted for assessment of suitability for infant feeding.

       Other groups of people who could potentially be adversely affected by water with high sulfate
concentrations are transient populations (i.e., tourists, hunters, students, and other temporary visitors)
and people moving to areas with high sulfate concentrations in drinking water from areas with low
sulfate concentrations.  This concern is based primarily on anecdotal reports rather than on published
studies. For example, an analysis of 300 responses to an informal survey conducted by the North
Dakota Department of Health suggested that water with sulfate levels  > 750 mg/L was considered
laxative by most consumers   (Peterson,  1951).  (Peterson noted that  a high  concentration of
magnesium sulfate was even more likely to have a laxative effect on consumers than was a high
concentration of sodium sulfate.) He also noted that reports of a laxative effect of water with a low
concentration of sulfate could have been from people new to the  area, whereas reports of no laxative
effect in areas of high sulfate concentration could have come  from people acclimated to sulfate
exposure.

       In another informal report, Moore (1952) evaluated data collected in North and South Dakota
on well water quality.  The data from  South Dakota included  67 wells with 1000 to 2000 mg/L
sulfate and indicated that the water was at least tolerable as drinking water with no apparent extensive
physiologic effect. The data from North Dakota included  information from 248 private drinking
water wells. As the concentration of sulfate in these wells increased, more adults reported a laxative
effect. For example, for well water containing <200 mg/L sulfate, only 22% of consumers reported
that their water had a laxative effect.  Water containing high (>  1000  mg/L) concentrations of
magnesium sulfate affected 62%  of consumers.   Neither of these reports  suggested  that the

                                          Page  5

-------
population affected considered the laxative effect of their drinking water to be  an adverse health
issue.

       Experimental studies of the association between exposure to sulfate and subsequent diarrhea
have been conducted in pigs and piglets and in human adults.  For example, groups of 10 artificially-
reared (using a mechanical "auto-sow") neonatal piglets were provided diets containing 0, 1200,
1600, or 2000 mg of added inorganic sulfate (as anhydrous sodium sulfate)/L of diet for 28 days
(Gomez  et al,  1995).   Sulfate concentrations of >1800  mg/L of diet caused  persistent, but
nonpathogenic diarrhea in the piglets. Growth was not affected in any of the exposure groups. A
study by Veenhuizen et al. (1992) found no adverse effect on nursery pig performance (e.g., mean
weight gain, feed consumption, water consumption, prevalence of diarrhea) when the pigs were fed
concentrations of up to 1,800 mg of sodium sulfate, magnesium sulfate, or a combination of sodium
and magnesium sulfate/L of diet for 16 or 18 days.

       In another study, Veenhuizen  (1993) conducted a water quality survey of 54 swine farms in
Ohio in which water samples were analyzed for concentrations of sulfates and total dissolved solids,
and found that sulfate concentrations ranged from 6  to 1600 mg/L.  There was  no association
between sulfate concentration and the prevalence of diarrhea on the farms.

       Heizer et al. (1997) provided four healthy adult subjects with drinking water containing
increasing levels of sulfate (0, 400, 600, 800, 1000, and 1200 mg/L from sodium  sulfate) for six
consecutive 2-day periods. In a single-dose study, six other volunteers received water with 0 or 1200
mg/L sulfate for two consecutive 6-day periods. In the dose-range study, there was a decrease in
mouth-to-anus appearance time (using colored markers) with increasing sulfate concentration. In the
single dose study, there was a significant increase in stool mass for the six days of exposure to sulfate
compared to the six days without exposure. None of the study subjects reported diarrhea.

       While the studies mentioned  above address the acute effects of sulfate on adult human
intestinal function and provide animal  data that can be extrapolated to people, it has not been feasible
to conduct an experimental study to verify the reported effects of exposure to high levels of sulfate
on human infants.  Esteban et al. (1997) conducted a field study in 19 South Dakota counties to
determine the risk for diarrhea in infants exposed to high levels of sulfate in tap water compared to
the risk for diarrhea in those unexposed. In this study, there was no significant association between
sulfate ingestion and the incidence of  diarrhea for the range of sulfate concentrations studied (mean
sulfate level 264 mg/L; range 0 - 2,787 mg/L). In addition, there was no dose-response or threshold
effect, and the results suggested that  breast milk has a more significant laxative effect than does
sulfate in drinking water. However, because the sample size  was small (274 infants) and the age of
the infants ranged from 6.5 weeks to 30 weeks, it is probable that some of the infants could have been
exposed (and become acclimated) to drinking water containing high levels of sulfate prior to being
enrolled in the study.
                                          Page  6

-------
Objective

       The objective of the present study was to provide additional information regarding whether
sensitive populations (infants and transients)  may be adversely affected by sudden exposure to
drinking water containing high levels of sulfate. Specifically, we designed a field investigation of
infants exposed to naturally occurring high levels of sulfate in the drinking water provided by public
water systems, and an experimental trial of exposure in adults.

Materials and Methods

       The protocols described in this report were reviewed and approved  by an expert panel
organized specifically to review this work and by the Institutional Review Boards (IRBs) of CDC,
Battelle Centers for Public Health Research and Evaluation, and the Texas Department of Health.

A.     Infants

       We attempted to conduct a prospective cohort study of newborn infants whose mothers did
not plan to breast-feed, but who planned to feed their infants formula mixed with tap water and who
planned to stay home with their infants for at least four weeks after giving birth. Unfortunately,
because we were not able to identify many new mothers who planned to feed their infants formula
mixed with tap water, we were unable to conduct the study. Below is a brief description of how we
planned to conduct this study.

Study Design

       We planned a prospective cohort study of infants born in geographic areas with naturally
occurring high levels of sulfate in the  drinking water provided by public water systems in New
Mexico, South Dakota, and Texas. Infants were to be enrolled at birth and followed for four weeks
to determine if there was an association between exposure to drinking water containing varying levels
of sulfate and reported cases of diarrhea.

       It may be difficult for  new mothers to identify diarrhea in their infants. Also, a new baby's
stools may be irregular in consistency for their first few days of life, making it difficult to identify a
"normal" elimination pattern.  To allow the babies' digestive systems time to stabilize and to prevent
the  infants from being immediately exposed to tap water, we planned to provide the mothers with
ready-to-feed formula.  The new mothers would be instructed to use this formula until the baby was
14 days old, and then to switch to powdered formula (that we would provide) that would be mixed
with their home tap water for the second two weeks of the study.

Sample size

       The baseline for diarrhea in infants from all causes is 13% (personal communication,  Center
for Infectious Disease, Centers for Disease Control and Prevention). With <* = 0.05 and 80% power,
a ratio of unexposed:exposed of 1:1, and assuming that the risk ratio for exposed compared with

                                          Page  7

-------
unexposed is approximately 2, we will needed a sample size of approximately 100 mothers and infants
per sulfate group. We planned to recruit approximately 110 mothers and infants per group to allow
for an anticipated 10% loss to follow-up.

Participant Recruitment

       For a woman to be included in our study, she had to meet the following criteria: (1) Be
accessible through a clinic that offers prenatal care, (2) Be 35 to 36 weeks pregnant (to exclude
severely premature infants and to be able to interact with the woman at her next prenatal visit), (3)
Be planning not to breast-feed her baby, (4) Be planning to feed her infant concentrated formula
mixed with her home tap water (which could be boiled), (5) Be planning to be at home with her infant
for at least one month, (6) Be living in a home (private house, apartment, mobile home, etc.) that was
served by a public water supply and that did not have a water filtration system, and (7) Be able to
read the study instruments in either English or Spanish.

       Using historic water quality data  for secondary drinking water  constituents (including
concentrations of sulfate, chloride, copper, fluoride, iron, manganese, and total dissolved solids)
provided by the states and Geographic Information System (GIS) techniques, we generated maps of
the geographic distribution of each water system that simultaneously identified the level of sulfate in
the water and the size of the  population each system served. Using the GIS maps, the number of
expected births in the participating states, the inclusion criteria we have described, and the assumption
that 25% of infants are not breast-fed, we planned a 6-month recruiting period. We planned to enroll
a maximum of 880 infants, 110 each from water systems within the following ranges of sulfate: < 250
mg/L (baseline or comparison group), 251-500 mg/L, 501-700 mg/L, with a maximum of 550 infants
from water systems  with sulfate levels greater than 701 mg/L.

Data collection

       Each mother would be instructed to complete a 4-week diary describing her baby's food
intake and elimination patterns beginning as soon as the baby was born and continuing until the end
of the baby's 4th week. In order to limit misclassification of diarrheal illness in  infants, we included
specific questions about the baby's stools rather than asking whether the baby had diarrhea.

       To verify the sulfate levels in a particular water system, we planned to request water samples
for each water system from which we recruited study participants. We planned to collect, store, ship,
and analyze the samples according to the methods described in the National Primary Drinking Water
Regulations—Sulfate; Proposed Rule (59 FR 65586) using state laboratories certified under EPA's
drinking water laboratory certification program.
                                          Page  8

-------
Statistical analysis

       We planned to examine the average number of episodes of diarrhea, as reported by the
mothers, in infants living in homes using tap water containing various levels of sulfate.  The
magnitude of the association between the consumption of sulfates in drinking water and the incidence
of diarrhea was to have been estimated by calculating the relative risk for diarrhea in each of the high-
sulfate groups  (251-500 mg/L, 501-700 mg/L,  701-1100 mg/L, and >1101 mg/L sulfate)  as
compared with the baseline group (0 - 250 mg/L sulfate).

Infant pilot study

       We conducted a pilot study of the planned recruitment methods and study instruments in four
counties in South Dakota. Local Public Health Nurses, who already had rapport with clinic patients,
were hired and trained to recruit study participants and conduct the activities associated with the
study.

       In the 4 counties (served by 3 Community Health Clinics which also provided the Women,
Infants, and Children [WIC] program services), 72 women were approached about participating in
the study during their prenatal visit to the public health clinics. Of these 72 women, 30 were ineligible
because they planned to breast-feed their infant, 23  were ineligible because they planned to use water
other than tap water to mix infant formula, and 11 were ineligible because they did not meet other
eligibility criteria.

       Of the eight women who were eligible to participate in the study, three refused, five agreed
to participate, and one completed all the study activities. Of the four women who did not complete
the study activities, two switched to bottled water as the source of water to mix with infant formula,
one chose to use ready-to-feed formula after two weeks of using the powdered formula, and the
fourth moved out of the area.

Self-administered questionnaire

       Because we experienced recruiting problems during the pilot study conducted in  South
Dakota, we developed a self-administered questionnaire (SAQ) made up of previously approved
questions from the infant study instruments we had planned to use to determine tap water use. We
provided English and  Spanish versions of the SAQ to 32 WIC clinics that offered prenatal care in
New Mexico, South Dakota, and Texas.  The clinics were located in areas with a range of sulfate
levels in the drinking water provided by the local public water systems

       The questionnaires were given to all women who came to the clinic during a two-week period,
and included questions about the source of their home tap water, what mothers of infants <3 months
old were currently feeding their babies, and how currently pregnant women planned to feed their new
infants.
                                         Page  9

-------
B.  Transient Populations

Study design

       To study the effects on adults of suddenly changing drinking water sources from one that has
little or no sulfate to one that is high in sulfate, we conducted an experimental study involving
volunteers from Atlanta, Georgia, including CDC employees and employees at the U.S. EPA Region
IV  office.  The sulfate concentration in the drinking water in Atlanta is very low,  thus  study
volunteers were not already acclimated to the presence of sulfate in their drinking water. Potential
volunteers were ineligible to participate in the study if they had traveled outside the U.S. within the
last two weeks before the interview, if they had any acute illnesses (e.g., flu or food-borne illness) or
if they had chronic medical problems that could affect bowel movements (e.g.,  irritable bowel
syndrome).

Sample Size

       The baseline for diarrhea in adults is 14% (i.e., at any point in time, 14% of adults will have
had diarrhea in the last month). With 95% confidence and 80% power, a ratio of unexposed: exposed
of 1:1 and assuming that the risk ratio for exposed compared with unexposed is approximately 4, we
will needed a sample size of 25 per sulfate group. To account for participant drop out during the
study and still have sufficient power to detect differences in reports of diarrhea among  the groups,
we attempted to recruit 150 volunteers, 30 each for the following sulfate exposure groups:  0 mg/L,
250 mg/L, 500 mg/L, 800 mg/L,  1200 mg/L.

Participant recruitment

       We attempted to recruit 150 volunteers from the National Center for Environmental Health
Campus and the Environmental Protection Agency Region 4 office in Atlanta, GA using signs posted
in building lobbies, a center-wide e-mail message, and personal requests.  Participation was very
limited for a number of reasons, including:  1) out of town travel,  2) concern about how having
diarrhea would affect work or evening activities, 3) personal distaste regarding the request to report
bowel movements, 4) concern about drinking "contaminated" water.

Data collection

       For this experiment, we added different levels of sulfate (from anhydrous sodium sulfate,
Sigma Chemical Company, St. Louis, MO) to bottled drinking water. The sodium salt was chosen
to be consistent with the studies conducted by Heizer et al. (1997) and to avoid the confounding that
could occur from exposure to magnesium sulfate (magnesium is also a laxative that might be present
in tap water).

        Volunteers were randomly assigned to one of 5 sulfate exposure groups (i.e., 0, 250, 500,
800, or  1200 mg/L sulfate from sodium sulfate) and were provided with bottled drinking water for

                                         Page  10

-------
6 days.  The bottled water for days 1, 2, and 6 contained plain water, and the bottles for days 3
through 5 contained the added sulfate.  The unfinished or empty bottles were returned and weighed
to determine how much water was consumed each day. Volunteers were blinded to the level of
sulfate in their drinking water.

       Study participants were also  requested to complete a short daily diary describing the
following:  any changes in smell, taste, or appearance of their drinking water compared to what they
normally drank, any visits to health care professionals, any medication they took for diarrhea, anyone
in their household who was currently ill with vomiting or diarrhea, any work or recreational activity
missed because of illness, and the number, size,  and consistency of their bowel movements.

       We conducted two trials of exposure to sulfate in drinking water.  In the first trial, the 63
participants who completed the study were divided among the dose groups as follows: 10 received
0 mg/L sulfate; 10 received 250 mg/L  sulfate; 10 received 500 mg/L  sulfate; 18 received 800 mg/L
sulfate; and 15 received 1200 mg/L sulfate. In the second trial, the 42 participants who completed
the study were divided among the dose groups as follows: 14 received 0 mg/L sulfate; 15 received
800 mg/L sulfate; and 13 received 1200 mg/L sulfate.

Statistical Analysis

       Using SAS statistical software (univariate analyses and logistic regression), we examined
water consumption, the relative frequency of diarrhea,  and the number and description of bowel
movements reported in the different dose groups.

Results

A.  Infant Study

       The 37 WIC clinics participating in the survey study draw their clients from 15 counties with
varying average levels of sulfate in the drinking water provided by local public water systems. Of the
1,388 SAQs completed by women attending the 32 public health clinics included in our study,  1184
were completed in  English and 204 in Spanish; one thousand one hundred and two (79%) women
reported that their home tap water came from a public water system,  and 143 (10%) had some type
of filter on their home tap. One thousand fifty three (76%) women reported that at least one person
in her household drinks water from the tap.

       Of the 951 pregnant women who completed the SAQ, 582 (61%) reported that they planned
to breast-feed their new infant and 141  (15%) reported that they planned to use infant formula mixed
with their home tap water. Of the 542  women with infants < 3 months old who completed the SAQ,
272 (50%)  reported that they had breast-fed their infant and 272 (50%) reported that they had used
infant formula mixed with their home tap water. In addition, 357 (66%) reported that they would not
use tap water to  mix  infant formula in the future.  A chi-square analysis showed that a pregnant
woman's plans to use tap water to mix infant formula was associated with the level of sulfate in her

                                        Page  11

-------
home tap water (%2 = 19.03, p < 0.001).  A chi-square analysis of the three lower doses combined
into one group and compared with the highest exposure group was also statistically significant ((/2
= 15.53, p< 0.001).

       To determine how many of the women who completed the SAQ would have been eligible to
participate in our study (based on the tap water source and use information), we examined the
responses of women who received their tap water from public water systems and who did not have
filters on their home tap. The result was that 403 women with infants < 3 months of age and 761
pregnant women met these eligibility criteria (Table 1).

       Among the 761 pregnant women who responded to the SAQ (and who received tap water
from a public water system and did not have a filter on their home tap), only 123 (16%) reported
planning to use tap water to mix formula for their infants. The largest numbers of women who had
used or were planning to use tap water to mix infant formula (and who received tap water from a
public water system and did not have a filter on their home tap) were in the 4 counties with average
sulfate levels < 250 mg/L.  Of the 365 pregnant women in areas with sulfate levels > 250 mg/L, 226
(62%) planned to use infant formula.  However, only 39(11%) planned to use infant formula mixed
with their tap water. Of the 183 women with infants <3 months old in areas with sulfate levels > 250
mg/L, 151 (83%) used infant formula, but only 35 (19%) reported having used infant formula mixed
with their tap water. Thus, of the 548 pregnant women and women with infants < 3 months old, only
74 infants had been or would be exposed to tap water containing > 250 mg/L sulfate.

       We examined hi storical drinking water analy si s data for the counties represented by the clinics
participating in the SAQ. In many areas where sulfate concentrations were high, the concentrations
of other constituents that affect the organoleptic quality of drinking water tended to be close to or
above the levels suggested by the National Secondary Drinking Water Standards (NSDWS) (40 CFR
143.3) (see Appendix). For example, in one county in South Dakota, the average levels  of drinking
water constituents that affect the taste of the water were: sulfate, 841 mg/L; total dissolved solids,
1462 mg/L (suggested level 500 mg/L); manganese 0.31 mg/L (suggested level 0.05 mg/L); iron, 0.59
mg/L (suggested level 0.3 mg/L); fluoride, 1.53 mg/L (suggested level 2.0  mg/L), and chloride, 125
mg/L (suggested level 250 mg/L).

B. Adult Study

       The results of the two sulfate trials were similar; therefore, we have reported the combined
results. One hundred five study participants were divided among the dose groups as follows:  24
received 0 mg/L sulfate; 10 received 250 mg/L sulfate; 10 received 500 mg/L sulfate; 33 received 800
mg/L sulfate; and 28 received 1200 mg/L sulfate.   The demographic information for the study
population was as follows: the mean age of participants was 42 years; the majority (62%) was female;
the races included in the study population were white (80%),
                                        Page 12

-------
Table 1:      Planned and previous use of tap water to mix infant formula by women who are
             currently pregnant and by women who have an infant < 3 months old (and who
             received their tap water from public water systems and who did not have filters on
             their home taps).
Mean Sulfate Level
(mg/L)
0 - 250 mg/L
251 -500 mg/L
501 -750 mg/L
>751 mg/L
TOTAL
Number of
Counties
4
5
3
3
15
Number
of
SAQs
682
455
135
116
1388
Women with Infants < 3 mos.
n
220
123
27
33
403
Formula Mixed
With Tap Water1
83 (38%)
20 (16%)
7 (26%)
8 (35%)
118(29%)
Pregnant Women
n
396
219
76
70
761
Formula Mixed
With Tap Water2
84 (21%)
17 (8%)
11(14%)
11(16%)
123 (16%)
1 Number of women with infants < 3 mos. of age who had used powdered formula mixed with water
and who used tap water.
2 Number of pregnant women planning to used powdered formula mixed with water and who are
planning to use tap water.
                                       Page  13

-------
black (13%), and Asian/Pacific Islander (7%).  Ninety-five percent of the participants were non-
Hispanic.

       During the study period, no one reported taking medication for diarrheal illness and no one
visited a health care professional for anything other than routine medical care.  In addition, no one
reported missing work or recreational activities because they had diarrhea.

       Because the presence of sulfate  affects the  organoleptic quality of drinking water,  we
examined reported differences in smell or taste of the water provided to study participants compared
to the water they normally drank. Among study participants who reported no difference in taste or
smell (compared to the water they normally drank) on days 1 and 2, the numbers of people who
reported a difference on days 3, 4, or 5 by dose group were: 4 (25%)  at 0 mg/L; 4  (57%) at 250
mg/L; 2 (50%) at 500 mg/L; 15 (79%) at 800 mg/L; and 14 (82%) at 1200 mg/L.

       We analyzed the number of bowel movements recorded each day by study participants, and
found that there were no statistically significant differences in the mean number of bowel movements
among the groups on days 3, 4, 5, or 6. Neither were there statistically significant differences in the
mean number of bowel movements reported when comparing days 1 and 2 (the days when there was
no sulfate in the water) with days 3, 4, and 5 within each dose group.

       Because  the  sulfate dose for each individual was dependent upon his  or her water
consumption, we examined the amount of water consumed by the different dose groups (see Figure
1).  Water consumption varied across the days of the study for all dose groups.  The average water
consumption in the 500, 800, and 1200 mg/L dose groups tended to decrease on days 3, 4, and 5
(when there was sulfate in the drinking water). All exposed groups consumed more water on day 6
(no sulfate in the drinking water) than on day 5 (the last exposure day).

       The frequencies of diarrhea reported by individuals exposed to varying levels of sulfate in their
drinking water are presented in Table 2.  Three different definitions of diarrhea were used: increase
in stool volume compared to normal, change in stool consistency to liquid or paste, and increase in
volume/change in consistency.   Because the effect of sulfate is dependent on both the amount of
water consumed and the weight of the participant (dose/kg of body weight), we  used logistic
regression to examine the reported frequency  of diarrhea (using the  three different definitions
described above) by sulfate dose ([concentration of sulfate in water x volume of water consumed on
days 3, 4,  and 5]/body weight).  The results of the logistic regression are presented  in Table 3.
Sulfate dose was not a statistically significant predictor of diarrhea in any of the models.

       To examine the data for a trend toward increased frequency of reports of  diarrhea with
increased sulfate intake, we included the dose as an ordinal variable in a logistic regression model of
osmotic diarrhea. There was no statistically significant increase in reports of diarrhea with increasing
dose (one-sided p = 0.099).
                                         Page  14

-------
Table 2.      Number and percent of study participants reporting diarrhea on days 3, 4, or 5 of the
             study (limited to those individuals who reported normal stool volume on days 1 and
             2 and who did not have family members who experienced vomiting or diarrhea on
             days 3, 4, or 5) following exposure to varying levels of sulfate in their drinking water.
Sulfate Dose
(mg/L)
0
250
500
800
1200
n
24
10
10
33
28
Osmotic Diarrhea1
No.4 (%)
2/18(11)
0/9
1/8 (12)
4/26(15)
5/27(18)
Diarrhea2
No.4 (%)
6/16 (38)
1/7 (14)
4/9 (44)
7/20 (35)
6/17 (35)
Diarrhea3
No.4 (%)
5/14 (36)
1/6 (17)
3/8 (38)
8/17 (47)
6/17 (35)
1 Reported as an increase in stool volume on days 3, 4, or 5 and limited to those who reported a
normal stool volume on days 1 and 2.

2 Reported as paste-like or liquid stools on days 3, 4, or 5 and limited to those who reported normal
stool consistency on days 1 and 2.

3 Reported as change in stool bulk or consistency on days 3,4, or 5 and limited to those who reported
normal stools on days 1 and 2.

4 (Number of people reporting diarrhea on days 3, 4, or 5)/(Number of people who reported normal
stools on days 1 and 2 and who did not report a family member ill with vomiting or diarrhea).
                                        Page 15

-------
Table 3.         Parameter estimates, standard error, and p values for logistic regression models
                 of diarrhea reported by study participants (limited to  those individuals who
                 reported normal  stool  volume on days 1 and 2 and who did not have family
                 members who experienced vomiting or diarrhea on days 3, 4, or 5) following
                 exposure to sulfate in drinking water.
Variables in Model
n
Model I1
Intercept 85
Dose
Model 22 67
Intercept
Dose
Model 33
Intercept 60
Dose
Parameter
Estimate

-1.825
4.629 E-7

-0.523
-1.64E-6

-0.529
1.415 E-6
Standard Error

0.430
7. 189 E-6

0.349
6.587 E-6

0.364
6.585 E-6
P

<0.01
0.94

0.13
0.80

0.15
0.83
1 Osmotic diarrhea reported as an increase in stool volume on days 3,4, or 5 and limited to those who
reported a normal stool volume on days 1 and 2.

2 Reported as paste-like or liquid stools on days 3, 4, or 5 and limited to those who reported normal
stool consistency on days 1 and 2.

3 Reported as change in stool bulk or consistency on days 3,4, or 5 and limited to those who reported
normal stools on days  1 and 2.
                                        Page  16

-------
We further examined the lowest three doses to determine if the response was statistically the same
across all three groups, and estimated a generalized linear model for the number of "successes" or
episodes of diarrhea in each of the lower sulfate dose groups (0, 250, and 500 mg/L). Since the
number of episodes of osmotic diarrhea was lowest in the 250 mg/L dose group, we performed a test
to determine whether that group truly differed from the 0 and 500 mg/L groups. We used a one-
tailed test to coincide with our alternative hypothesis that the number of episodes of osmotic diarrhea
was lower in the 250 mg/L dose group than in the other two dose groups. The resulting p-value was
.09, suggesting a flat dose response at those sulfate levels.

    When the lowest doses (0, 250, and 500 mg/L) were collapsed into one group,  3 (9%, n = 35)
individuals reported an increase in stool volume, 34 (33%, n = 103) individuals reported a change in
stool consistency, and 11 (34%, n = 32)  reported a change in  stool volume or consistency (on days
3, 4, or 5, with no report of household members having diarrhea or vomiting).  Compared with the
data for the 800 mg/L and 1200 mg/L doses presented in Table 2, there was an increase in reports
of osmotic diarrhea with increasing dose.

Discussion

    The purpose of these studies was to provide  information about the health effects (primarily
osmotic diarrhea) in sensitive populations (infants and transients) of exposure to high levels of sulfate
in drinking water. The most efficient method available to examine this question would be to conduct
an experimental trial in which a subj ect consumes a known amount of sulfate in water and reports any
subsequent cases of osmotic diarrhea. However, there are a number of difficulties associated with
conducting this type of study.  First, it is  difficult for study participants to identify osmotic diarrhea.
Second, while there is some agreement about a baseline incidence of diarrhea in adults, it is likely to
be under-reported and there is no gold standard with which to compare our control  group.  Third,
it is difficult for study participants to consume a large volume of water containing very high levels of
sulfate because of the adverse effect sulfate has on the organoleptic quality of the water. Finally, it
would not be ethical to conduct an experimental trial involving exposure of infants to a substance that
may cause diarrhea and subsequent dehydration.

    For the infant study, we planned to conduct a prospective  cohort study of newborn infants. We
planned to identify highly exposed populations  and recruit pregnant women living  in the relevant
geographic areas.  We anticipated that approximately 75% of  women would be ineligible to
participate because they were planning to breast-feed their infants.  We also anticipated that, of those
25% of women who did not plan to breast-feed, 50 % would not be eligible (either because their tap
water came from a source other than a public water system or because they had a filter on their home
tap) or would refuse to participate.

    During the pilot study conducted in South Dakota, only eight of 72 (11%) women were eligible
to participate.  Of the five women (63% percent of those eligible) who agreed to participate, two
decided to use bottled water to mix infant formula and one decided to use ready-to-feed formula after
starting the study.  Thus, plans to breast-feed, to use ready-to-feed  formula, or to use water other
than tap water prevented 67 (93%)  of the women from being eligible to enter or complete the study.
                                         Page  17

-------
   Most (74%) of the women approached during the pilot study were not eligible to participate
because either they planned to breast-feed their new infant or they did not plan to use their tap water
to mix infant formula.   This finding suggested that very few infants were actually exposed to tap
water containing high levels of sulfate. In order to gain more specific information concerning the
number of infants actually exposed, we developed a self-administered  questionnaire (SAQ) to
determine how many women planned to use tap water to mix infant formula for their new babies.

   The results from the SAQ were that more than half (61%) pregnant women reported planning to
breast-feed their infants. In addition, of those who planned to use infant formula mixed with water,
most (84%) planned to use water other than tap water.  These results were consistent with our
findings from the pilot study and indicated that only a very small number of women who live in areas
with high levels of sulfate in the tap water provided by public water systems planned to give this
water to their infants.

   The geographic areas where there were high levels of sulfate in the public water systems tended
to be rural areas that were sparsely populated. In addition, the historical water quality data indicated
that many of the geographic areas with high levels of sulfate in the water provided by public water
systems also had high levels of other constituents that adversely affect water quality (see Appendix).
Our SAQ results indicated that a very small number of women living in these areas planned to give
their babies formula mixed with tap water.  Thus, it would not be practical to conduct a study of the
effects of sulfate on diarrhea in infants by recruiting a study population exposed to naturally occurring
high levels of sulfate in the drinking water provided by public water systems. However, because there
is a concern about exposing young  infants to tap water containing  high levels of ions, including
sulfates, a conservative approach would be to recommend that individuals using their tap water to
feed infants have their water tested for sulfate content or choose an alternative water source known
to be low in sulfate.

   For the adult study, we conducted a  modified  experimental trial.   Study participants were
randomly  assigned to a sulfate  dose  level and were blinded  to the dose they  were  receiving.
However, at the high levels of sulfate, study participants could taste the  change in their drinking
water, and so could tell that they were in an exposed group. It is not clear whether, or in which
direction, this knowledge might bias the responses provided by the study participants. It is interesting
to note that some of the participants who were in the control group complained that the water tasted
bad and they weren't sure they would be able to complete the study.

   Given the limitations of the study design, we examined the incidence of diarrhea in study
participants exposed to different levels of sulfate in their drinking water in two ways. First, we
examined the number of reports of diarrhea across exposure groups using three different definitions
of diarrhea (see Table 2).  There was an increase in the number of people who reported osmotic
diarrhea in the most highly exposed groups (800 mg/L and 1200 mg/L) compared to the controls (0
mg/L), but the differences were not statistically significant.  There were  no associations between
sulfate dose and the number of reports of diarrhea when the other two  definitions of diarrhea were
used.  These results are consistent with the results on human subjects reported by Heizer et. al.
(1997) (i.e., water containing 1200 mg/L sulfate from sodium sulfate produced mild increases in stool
weight, decreases in self-reported stool consistency, but no  complaints of diarrhea).

                                         Page  18

-------
   Because water consumption varied for each individual in the study, we also examined the sulfate
dose as the amount of sulfate consumed per kg of body weight. We used logistic regression analysis
to examine the association between sulfate dose (mg/kg body weight) and reports of diarrhea using
the three definitions (see Table 3). There were no statistically significant dose-response associations
between sulfate dose and reports of diarrhea.

   The study participants exposed to 250 mg/L sulfate actually reported fewer incidents of diarrhea
than the control group (0  mg/L sulfate) did. Similar to the conclusions drawn by Heizer, et. al,
(1997) and Chi en et al. (1968), our conclusion is that it is unlikely that exposure to sulfate in drinking
water at concentrations below 600 mg/L would cause diarrhea in people.  Thus, the 0 mg/L, 250
mg/L, and 500 mg/L exposures may all be below a threshold dose required to produce a biological
effect, and, if so, a dose-response effect within this low range would not be expected.

   Study participants were not requested to drink a specified amount of water.  Because we were
concerned that the changes in the organoleptic quality of the water would result in lower consumption
among people in the higher dose groups,  we examined water consumption (see Figure 1).  The
average water consumption decreased during the exposure period (days 3-5) for all exposure groups.
There was no pattern in the average water consumption by the control group over the 6-day study
period.

   We were not able to conduct the dose-response studies that were requested as part of the Safe
Drinking Water Act Amendments of 1996.  Instead, we convened an expert  workshop whose
members reviewed the available literature and addressed a series of questions about the health effects
from exposure to sulfate in drinking water:  1) Do reported studies suggest that a certain sulfate level
would not be likely to cause adverse effects?, 2) Does the literature support acclimatization or
adaptation?, 3) Can an infant dose-response study be done anywhere in the U.S.  or Canada?, 4) Is
there enough scientific evidence of adverse health effects from sulfate in drinking water to support
regulation? The workshop was held in Atlanta, Georgia on September 28, 1998.  The workshop
report accompanies this document.

Conclusion

   The purpose of this project was to examine the association between consumption of tap water
containing high levels of sulfate and reports of osmotic diarrhea in susceptible populations (infants
and transients, i.e., those acutely exposed). We were unable to conduct a study of infants because
we could not identify enough exposed individuals from which to draw a study population. The results
of our SAQ examining tap water use indicated that most pregnant women who completed the survey
planned to breast-feed their infants. Of those who planned to use formula mixed with water, most
did not plan to use tap water to mix the formula. In our experimental trials with adult volunteers, we
did not find a significant dose-response association between acute exposure to sodium sulfate in water
(up to 1200 mg/L) and reports of diarrhea. However, we did find a weak (not statistically significant)
increase in reports of diarrhea at the highest dose level when it was compared to the combined lower
doses.
                                         Page  19

-------
                                       References

Chien L, H Robertson, J. Gerrard.  Infantile gastroenteritis due to water with high sulfate
content. Can Med Assoc J 1968;99:102-104.

Esteban E, Rubin C, McGeehin M, Flanders D, Baker MJ, Sinks TH.  Evaluation of human health
effects associated with elevated levels of sulfate in drinking water: a cohort investigation in South
Dakota, manuscript submitted. Int J Occup Med Environ Health. 1997;3:171-176.

Gomez GG, Sandier RS, and Seal E Jr.  High levels of inorganic sulfate cause diarrhea in neonatal
piglets. JNutr. 1995;125:2325-2322.

Heizer WD, Sandier RS, Seal E, Jr., Murray SC, Busby MG, Schliebe BG, Pusek SN.  Intestinal
effects of sulfate on drinking water on normal human subjects.  Dig Dis Sci  1997;42 (No.
5):1055-1061.

Moore FW.  Physiological effects of the consumption of saline drinking water. A progress report
to the 16th Meeting of the Subcommittee on Sanitary Engineering and Environment. Appendix
B. January 1952.  Washington, D.C.: National  Academy  of Sciences, 1952.

Peterson NL. Sulfates in drinking water.  Official Bulletin: North Dakota Water and Sewage
Works Conference.  1951; 18:11.

U.S. Environmental Protection Agency. Fact Sheet: National Secondary Drinking Water
Standards. Office of Water.  EPA 570/9-91-019FS. 1991.

U.S. Environmental Protection Agency. National Primary Drinking Water Regulations—Sulfate;
Proposed Rule (59 FR 65586).  December 20, 1994 Federal Register notice.

Veenhuizen MF, Shurson GC, Kohler EM.  Effect of concentration and source of sulfate on
nursery pig performance and health.  JAVMA.  1992;201(No. 8): 1203-8.

Veenhuizen MF. Association between water sulfate and dirrhea in swine on Ohio farms.
JAVMA. 1993;8:1255-60.
                                        Page  21

-------
     Figure 1: Mean volume of water in mL consumed by volunteers in the two adult trials on each
              day of the study.
      2000
T3
CD
3
CO
c
o
O
03
O
CD
E
_3

§
C
CO
CD
1500
1000  -
 500  -
 Sulfate Dose Group (n)
    Omg/L(24)
   250mg/L(10)
   500mg/L(10)
B 800mg/L(32)
   1200mg/L(28)
                                            Day of Study
                                          Page 22

-------
                            Appendix

Organoleptic Secondary Drinking Water Constituents and Tap Water Use
                    for Infant Formula: Summary
                             Page 23

-------
Organoleptic Secondary Drinking Water Constituents and Tap Water Use for Infant Formula: Summary
County
(No. PWS)
Drinking Water Constituents: Mean and Range
Chloride
MCL1:
250 mg/L
Copper
MCL:
Img/L
Fluoride
MCL:
2.0 mg/L
Iron
MCL:
0.3 mg/L
Manganese
MCL:
0.05 mg/L
Sulfate
MCL:
250 mg/L
IDS2
MCL:
500 mg/L
No.
Clinics
No.
SAQs3
Eng/Sp
Women with Infants (< 3 mos.)
n
Powdered Formula
n
Tap Water4
Pregnant Women
n
Breast-Feed
n
Powdered Formula
n
Tap Water5
New Mexico
Bernalillo
(28)
Chaves (6)
Dona Ana
Edd (8)
McKinley
(4)
Otero (13)
Socorro (2)
44
0-225
42
12-35

102
5-467
46.5
28-65.1
199
37-356
51.2







0.54
0.1-2.97
1.08
0.73-1.90

0.63
0.22-0.83
1.49
1.25-2
0.36
0.15-0.78
0.5
0.09
0-0.1
0.10
(lvalue)


0.11
0.02-0.2

0
0.05
0.05-0.05
0.08
(1 value)


0.04
0.02-0.05

0
116
19-360
318
46-462
165
16-283
328
41-687
789
224-2187
506
146-820
899
96-1702
481
102-1000



760
(1 value)


8
1
1
2
1
1
1
112/71
81/9
18/20
98/9
33/3
65/2
31/6
48
19
13
41
10
14
10
26
17
8
36
6
10
8
14(29%)
1 (6%)
0
7 (19%)
4 (40%)
3 (30%)
0 (0%)
132
56
26
55
19
33
26
94
28
14
30
13
20
18
47
35
12
31
9
17
15
14(11%)
2 (4%)
0
1 (2%)
3 (16%)
2 (6%)
5 (19%)
South Dakota (Data from treated [T] water samples if that data was provided, otherwise average of other types of samples [R=raw, RT=raw treated, RC=raw composite, RTC=raw treated composite with limited treatment])
Beadle (10)
Brown (8)
Spink (5)
City of
Pierre
(5)
City of
Rapid City
(13)
125
21-353
87
32-235
161
6-476
96
11-213
8
2-26





1.53
0.72-2.58
0.85
0.16-1.64
2.07
0.06-5.17
0.65
0.25-2.03
0.47
0.17-1.42
0.59
0.1-1.6
0.13
0.0-0.3
0.6
0.1-2.1
0.69
0.2-2.0
0.1
0.1-0.3
0.31
0.02-2.26
0.18
0.02-0.63
0.23
0.02-0.45
1.33
0.02-3.48
0.025
0.02-0.04
841
193-1255
409
171-1101
737
10-1213
528
235-1105
96
14-433
1463
509-2093
1011
486-1849
1608
36-2794
1304
462-2032
324
142-697
1
1
1
1
1
43/0
33/0
24/0
44/0
325/0
13
3
6
7
99
10
3
6
5
80
4 (40%)
3 (100%)
4 (67%)
0 (0%)
54 (68%)
25
26
14
29
173
12
12
11
15
112
15
21
8
19
86
3 (12%)
1 1 (42%)
3 (21%)
6(21%)
49 (27%)
                     Page  24

-------
Texas
Midland/
Ector(14)

Hays (8)


Travis (1)

473
216-1386

32
16-73

61

0.011
0.006-
0.04
0.014
0.006-
0.039
0.006

1.67
0.8-3.75

2.2
0.3-3.1

0.6

0.06
0.01-0.33

0.11
0.02-0.48

0.02

0.006
0.002-
0.008
0.050
0.002-0.35

0.008

450
277-916

534
38-1648

46

1606
950-3381

1030
322-2697

243

4


2


11

147/40


33/2


97/39

45


15


60

38


12


35

4(11%)


5 (42%)


15(43%)

72


11


64
761
24


10


44
457
54


2


38
409
3 (4%)


0 (0%)


18(%)
123
'Maximum Contaminant Level; U.S. Environmental Protection Agency Secondary Drinking Water Standards.
2Total Dissolved Solids
3Number of completed Self-Administered Questionnaires (SAQs)
4Number of women using tap water (PWS or community well) to mix infant formula for their infants <3 months of age
5Number of pregnant women planning to use tap water (PWS or community well) to mix infant formula
. Data not available
                                                           Page  25

-------