United
States
Office
of
Water
EPA­
822­
R­
00­
008
Environmental
Protection
4301
April
2000
Agency
ESTIMATED
PER
CAPITA
WATER
INGESTION
IN
THE
UNITED
STATES
April,
2000
Based
on
Data
Collected
by
the
United
States
Department
of
Agriculture's
1994
 
96
Continuing
Survey
of
Food
Intakes
by
Individuals
Support
provided
under
EPA
Contracts:
#
68
 
C4
 
0046
and
#
68
 
C
 
99
 
233

ACKNOWLEDGMENTS
Authors
Helen
L.
Jacobs,
M.
S.
U.
S.
EPA,
Office
of
Science
and
Technology
Julie
T.
Du,
Ph.
D.
U.
S.
EPA,
Office
of
Science
and
Technology
Henry
D.
Kahn,
D.
Sc.
U.
S.
EPA,
Office
of
Science
and
Technology
Kathleen
A.
Stralka,
M.
S.
Science
Applications
International
Corporation
Technical
Support
Rita
Schoeny,
Ph.
D.
U.
S.
EPA,
Office
of
Science
and
Technology
Carol
Lang,
M.
S.
U.
S.
EPA,
Office
of
Pesticide
Programs
Annetta
Cook,
M.
S.
U.
S.
Department
of
Agriculture
Dung
Bich
Phan,
B.
S.
Science
Applications
International
Corporation
Jan
Coulson,
B.
S.
Science
Applications
International
Corporation
Lynn
Donaldson,
M.
S.
Science
Applications
International
Corporation
Document
Review
Denis
Borum,
B.
S.
U.
S.
EPA,
Office
of
Science
and
Technology
Joyce
Donohue,
Ph.
D.
U.
S.
EPA,
Office
of
Science
and
Technology
Management
Support
JeanetteWiltse,
Ph.
D.,
J.
D.
Director,
Health
and
Ecological
Criteria
Division
U.
S.
EPA,
Office
of
Science
and
Technology
Sheila
Frace,
B.
S.
Director,
Engineering
and
Analysis
Division
U.
S.
EPA,
Office
of
Science
and
Technology
ii
SAB
REVIEW
PANEL
MEMBERS
AND
FEDERAL
EXPERT
The
following
professionals
provided
technical
comments
and
insight
regarding
the
content
and
technical
approach
of
the
efforts
leading
to
this
report.
The
listed
individuals
are
members
of
the
Drinking
Water
Intake
Subcommittee
to
the
Environmental
Protection
Agency's
Science
Advisory
Board
(
SAB).
Mr.

Thomas
O.
Miller
is
the
Designated
Federal
Official
for
the
EPA.

SAB
Review
Co
 
Chairs:

Dr.
Henry
A.
Anderson,
Chief
Medical
Officer,
Wisconsin
Bureau
of
Public
Health,
Madison,
WI
Dr.
Richard
Bull,
Senior
Staff
Scientist,
Battelle
Pacific
Northwest
Laboratories,
Richland,
WA
SAB
Review
Panelists:

Dr.
Judy
Bean,
Director,
Biostatistics
Program,
Children's
Hospital
Medical
Center,
Inc.,
Cincinnati,
OH
Dr.
Cynthia
Bearer,
Assistant
Professor,
Case
Western
Reserve
University,
Cleveland,
OH
Dr.
John
Evans,
Program
in
Environmental
Science
and
Risk
Management,
Harvard
School
of
Public
Health,
Cambridge,
MA
Dr.
Anna
Fan
 
Cheuk,
Chief,
Pesticide
and
Environmental
Toxicology
Section,
Office
of
Environmental
Health
Hazard
Assessment,
California
Environmental
Protection
Agency,
Oakland,
CA
Dr.
Richard
Gilbert,
Staff
Scientist,
BattelleWashington
Office,
Washington,
D.
C.

Dr.
Barbara
L.
Harper,
Toxicologist,
Yakama
Indian
Nation,
Richland
WA
Dr.
Michael
Jaycock,
Senior
Research
Fellow,
Rohm
and
Haas
Co.,
Spring
House,
PA
Dr.
Kai
 
Shen
Liu,
Epidemiologist,
California
Department
of
Health
Services,
Berkeley,
CA
Dr.
Edo
Pellizzari,
Vice
President
for
Research,
Research
Triangle
Institute,
Research
Triangle
Park,
NC
Dr.
Barbara
Peterson,
President,
Novigen
Sciences,
Inc.,
Washington,
D.
C.

Federal
Expert:

Ms.
Sharon
Mickel,
Nutritionist,
Food
Surveys
Research
Group,
United
States
Department
of
Agriculture
iii
PREFACE
This
report
presents
current
estimates
of
per
capita
water
ingestion.
The
basis
for
these
estimates
is
dietary
and
demographic
data
collected
during
a
1994
through
1996
survey
conducted
by
the
United
States
Department
of
Agriculture
(
USDA).
In
this
survey,
known
as
the
Continuing
Survey
of
Food
Intakes
by
Individuals
(
CSFII),
two
non
 
consecutive
days
of
food
ingestion
data
were
collected
from
a
sample
of
more
than
15,000
individuals
in
the
50
United
States
and
the
District
of
Columbia.
Respondent
information,
in
conjunction
with
food
code,
recipe,
and
nutrient
data
from
USDA,
forms
the
means
of
estimating
per
capita
ingestion
of
plain
drinking
water
(
direct
water)
and
water
ingested
indirectly.
Water
used
in
the
final
preparation
of
foods
and
beverages
at
home,
or
by
food
service
establishments
such
as
school
cafeterias
and
restaurants
is
defined
as
indirect
water.
Quantities
of
ingested
water
reported
in
the
USDA
1994
through
1996
CSFII
are
averaged
by
participant
to
generate
a
two
 
day
average.
These
daily
average
ingestion
amounts
comprise
the
empirical
distributions
from
which
mean
and
upper
percentile
per
capita
ingestion
estimates
are
produced.

The
CSFII
survey,
and
consequently
estimates
reported
in
this
document,
extend
to
the
population
of
the
United
States.
We
augment
population
per
capita
ingestion
estimates
with
estimates
of
per
capita
ingestion
by
various
population
subsets.
These
population
subsets
include
(
1)
gender
and
age
categories
and
(
2)
pregnant,
lactating,
and
childbearing
 
age
women.
Subpopulation
ingestion
estimates
support
assessments
of
"
at
risk"
populations.

EPA
generated
the
estimates
in
this
report
in
response
to
legislative
mandates
in
Safe
Drinking
Water
Act
Amendments
of
1996.
These
mandates
require
up
 
to
 
date
information
on
water
ingestion
to
identify
subpopulations
at
elevated
risk
of
health
effects
from
exposure
to
contaminants
in
drinking
water.
These
up
 
to
 
date
estimates
also
support
characterization
of
health
risks
to
sensitive
populations
from
contaminants
in
drinking
water.
The
estimates
in
this
document
characterize
the
empirical
distributions
of
two
 
day
average
per
capita
ingestion
of
water
for
specific
subpopulations.
Subpopulation
estimates
apply
to
demographic
categories
but
do
not
distinguish
individuals
with
a
history
of
serious
illness
or
with
lifestyles
that
effect
water
consumption.

Water
ingestion
rates
for
the
overall
population
and
for
subpopulations
have
several
important
applications
within
the
United
States
Environmental
Protection
Agency
(
USEPA).
Information
on
water
intake
is
used
in
risk
assessment
and
regulations
which
involve
default
values
for
water
ingestion
and
in
the
estimation
of
risks
to
highly
exposed
and/
or
sensitive
populations.

It
is
important
to
emphasize
that
risk
is
a
function
of
both
exposure
and
sensitivity.
Sensitivity
is
determined
by
genetics,
developmental
stage
(
old
as
well
as
young),
lifestyle,
and
preexisting
disease
iv
conditions.
With
the
exception
of
age,
these
other
determinants
of
sensitivity
are
not
addressed
in
this
report.

Water
ingestion
estimates
in
this
document
support
the
evaluation
and
possible
revision
of
the
standard
water
ingestion
quantities
of
two
liters
for
a
70
 
kilogram
average
adult
and
one
liter
for
a
10
 
kilogram
child.
These
standard
quantities
are
used
by
many
federal
agencies
including
the
EPA
and
theWorld
Health
Organization
(
WHO).
The
two
liter
standard
is
supported
by
a
1989
National
Cancer
Institute
report
on
tap
water
consumption
(
Ershow
and
Cantor,
1989).
Estimates
of
water
ingested
in
this
report
differ
from
the
estimates
reported
by
Ershow
and
Cantor
for
several
reasons.
Ershow
and
Cantor's
estimates
are
based
on
data
from
the
1977
 
78
USDA
National
Food
Consumption
Survey
while
the
estimates
reported
here
are
based
on
1994
 
96
data.
Also,
the
1989
report
presents
estimates
of
tap
water
ingestion.
Ershow
and
Cantor
define
tap
water
as
"
water
from
the
household
tap."
In
this
report,
water
coming
from
the
tap
is
distinguished
by
source.
Sources
of
water
coming
from
the
tap
may
include:

community
water,
household
well
or
cistern,
a
household
or
public
spring,
and
other.
Thus,
estimates
in
this
report
are
expected
to
differ
from
those
reported
in
1989
because
the
estimates
in
this
report
incorporate
more
recent
ingestion
data
and
thus
reflect
changes
in
ingestion
behavior.
Also,
estimates
will
differ
between
the
1989
report
and
this
report
because
the
sources
of
water
ingested
are
more
definitive
in
this
report.
A
third
way
that
the
estimates
in
this
report
differ
from
those
in
the
1989
report
is
that
the
1994
 
96
data
include
water
ingestion
by
pregnant
and
lactating
women.
These
women
were
excluded
from
the
1989
report.
To
further
address
changes
in
water
ingestion
patterns,
this
report
provides
separate
estimates
for
community
water,
bottled
water,
and
water
from
other
sources.

This
report
consists
of
the
following
chapters:

°
Executive
Summary,
summarizes
the
most
pertinent
information
contained
in
this
report,

including
the
main
features
of
the
CSFII
data
collection
and
results
from
the
analysis.

°
Chapter
1,
Definitions,
identifies
water
 
related
terms
used
in
the
report.
Definitions
distinguish
indirect
water
from
direct
water
and
identify
water
sources.

°
Chapter
2,
Sources
of
Data,
describes
the
surveys;
summarizes
the
method
of
data
collection;
and
identifies
the
respondent
data
files
and
concomitant
information
files
used
to
establish
the
estimates.
Appendix
D
provides
the
details
of
the
sample
design.

°
Chapter
3,
Methods,
presents
the
means
of
determining
source
and
amount
of
direct
water
ingested
by
survey
respondents.
Conventions
for
identifying
and
determining
the
amount
of
water
ingested
indirectly
through
food
preparations
are
also
presented.
Data
convention
descriptions
are
followed
by
a
summary
of
the
statistical
methods
used
for
generating
mean
and
empirical
percentile
v
estimates
and
the
size
of
the
subpopulation
to
which
the
estimates
are
applicable.
Appendix
D
records
statistical
estimation
formulae.

°
Chapter
4,
Results,
provides
an
overview
of
key
results.
These
results
are
augmented
with
graphical
presentations
and
numerous
tables
of
the
empirical
distribution
of
estimated
average
daily
per
capita
ingestion
of
water.

°
Chapter
5,
Discussion,
discusses
the
advantages
and
disadvantages
of
the
CSFII
for
estimating
per
capita
water
ingestion
in
the
United
States.
Sources
of
error,
bias,
and
uncertainty
are
defined,

and
the
report's
conclusions
are
presented.

Material
included
in
the
appendices
augment
the
data
convention
descriptions
and
methods
described
in
Chapters
2
and
3.
Appendix
E
presents
tabulated
estimates
of
per
capita
water
ingestion
by
water
source
and
subpopulation
for
all
respondents
and
for
"
consumers
only."

°
Appendix
A,
CSFII
Survey
Questions
Pertaining
to
Water
Ingestion,
lists
the
household
level
questions
that
are
used
to
determine
water
source,
sample
person
questions
that
identify
the
number
of
fluid
ounces
and
source
of
directly
ingested
water,
and
food
item
questions
for
determining
foods
with
water
added
at
home
or
by
a
food
service
facility.

°
Appendix
B,
Examples
of
Procedures
Used
in
the
Estimation
of
Indirect
Water
Ingestion,

provides
three
sets
of
examples.
For
food
codes
that
were
prepared
at
home
or
by
food
service
establishments,
Appendix
B1
identifies
how
the
proportion
of
indirect
water
in
100
grams
of
each
food
was
estimated
and
provides
examples.
Appendix
B2
provides
examples,
supplied
by
USDA,

of
how
to
estimate
preparation
water
absorbed
in
foods
such
as
cooked
pasta,
rice,
cereal
grains,

beans,
and
legumes.
USDA
guidance
and
examples
for
calculating
the
percent
and
amount
of
moisture
in
100
grams
of
food
follow
in
Appendix
B3.

°
Appendix
C,
1994
 
96
CSFII
Food
Codes,
lists
CSFII
Food
Codes
at
the
three
 
digit
level
and
the
assignments
of
percentage
of
indirect
water
and
commercial
water
in
C1
and
C2,
respectively.

Commercially
added
waters
are
not
included
in
the
ingestion
estimates
presented
in
this
report.

Appendix
C3
lists
food
codes
and
their
corresponding
proportions
of
water
in
100
grams
of
food.

°
Appendix
D,
Statistical
Methods
and
Sample
Design,
provides
the
statistical
formulae
for
generating
point
and
interval
estimates
about
the
mean
and
upper
percentiles
of
the
distribution
of
two
 
day
average
per
capita
water
ingestion.
This
appendix
also
provides
the
details
of
the
sample
design.
vi

Appendix
E,
Per
CapitaWater
Ingestion
Estimates,
includes
tabulated
presentations
of
per
capita
water
ingestion
estimates.
All
estimates
are
from
empirical
distributions
of
two
 
day
average
amounts
of
water
ingested.
This
appendix
presents
tables
for
the
entire
population
and
for
individuals
in
specific
subpopulations
in
four
parts.
Parts
I
and
II
record
estimates
of
direct,

indirect,
and
both
direct
and
indirect
water
ingestion
for
all
individuals.
Parts
III
and
IV
contain
water
ingestion
estimates
for
"
consumers
only."
These
estimates
only
include
individuals
who
reported
ingestion
of
the
water
under
consideration.
Therefore,
these
estimates
do
not
include
individuals
who
reported
zero
amounts
of
water
ingested
from
the
water
source
under
consideration.
Biological
and
commercially
added
waters
are
not
included
in
the
amounts
of
indirect
water
ingested.

Five
sets
of
estimates
comprise
each
part
of
this
appendix.
The
five
sets
differ
by
the
source
of
water
ingested.
These
sources
are
community
water,
bottled
water,
water
from
other
sources,

missing
source,
and
all
sources.
Each
part
contains
three
tables
of
estimates
for
each
water
source.

These
tables
report
water
ingestion
estimates
by
gender
and
broad
age
category;
fine
age
category;

and
pregnant,
lactating,
and
childbearing
 
age
women.
For
each
water
source,
ingestion
estimates
contained
in
Parts
I
and
Parts
III
are
reported
in
units
of
milliliters/
person/
day.
Units
for
Parts
II
and
IV
are
in
milliliters/
kilogram
of
body
weight/
day.


Appendix
F,
Final
SAB
Report
and
EPA
Response,
includes
the
results
of
a
review
of
the
July
1999
version
of
this
report
by
the
Drinking
Water
Intake
Subcommittee
(
DWIS),
a
special
subcommittee
of
the
EPA
SAB.
The
EPA's
response
to
this
report
is
also
included.
vii
THIS
PAGE
INTENTIONALLY
LEFT
BLANK
1For
the
purpose
of
this
report,
indirect
water
does
not
include
water
found
naturally
in
foods
(
biological
water)
and
water
added
by
commercial
food
and
beverage
manufacturers
(
commercial
water).

2References
in
this
report
to
the
ingestion
of
community
water,
bottled
water,
and
other
water
refer
to
the
ingestion
of
the
combined
amount
of
direct
and
indirect
community,
bottled,
or
other
water,
respectively.

viii
EXECUTIVE
SUMMARY
The
objective
of
the
report
is
to
provide
current
estimates
of
water
ingestion
for
the
population
of
the
United
States
and
selected
subpopulations.
The
subpopulations
include
gender
and
age
categories,

pregnant
women,
lactating
women
and
women
of
childbearing
age.
These
ingestion
estimates
may
be
used
in
estimating
risk
to
human
health
from
the
ingestion
of
contaminated
waters.
Knowledge
of
water
ingestion
is
of
fundamental
importance
to
the
mission
of
the
Office
of
Water,
and
credible
national
estimates
are
of
great
utility
to
many
EPA
programs.
In
particular,
the
estimates
support
the
development
of
risk
assessments
based
on
the
ingestion
of
water
that
may
be
contaminated.
The
Safe
Drinking
Water
Act
Amendments
of
1996
require
EPA
to
identify
subpopulations
at
elevated
risk
of
health
effects
from
exposure
to
contaminants
in
drinking
water
and
to
conduct
studies
characterizing
health
risk
to
sensitive
populations
from
contaminants
in
drinking
water.
The
process
of
establishing
human
risk
requires
up
 
to
 
date
information
on
water
ingestion
and
this
report
responds
to
that
need.

The
reported
estimates
were
calculated
using
data
from
the
combined
1994,
1995,
and
1996
Continuing
Survey
of
Food
Intakes
by
Individuals
(
CSFII),
conducted
by
the
United
States
Department
of
Agriculture
(
USDA).
The
CSFII
is
a
complex,
multistage
area
probability
sample
of
the
entire
United
States
and
is
conducted
to
survey
the
food
and
beverage
intake
of
the
United
States.
The
CSFII
collected
two
non
 
consecutive
days
of
food
ingestion
data
from
a
sample
of
more
than
15,000
individuals.
The
two
days
of
dietary
intake,
in
conjunction
with
food
code,
recipe,
and
nutrient
data
from
the
USDA,
were
used
to
identify
the
direct
(
plain
drinking
water)
and
indirect
water
consumed
by
each
respondent.
Indirect
water
is
defined
as
water
used
in
the
final
preparation
of
foods
and
beverages
at
home,
or
by
food
service
establishments
such
as
school
cafeterias
and
restaurants.
Quantities
of
ingested
water
reported
were
averaged
by
participant
to
generate
a
two
 
day
average.
These
daily
average
ingestion
amounts
comprise
the
empirical
distributions
from
which
mean
and
percentile
per
capita
ingestion
estimates
are
produced.

This
report
provides
ingestion
estimates
of
direct
water,
indirect
water
and
both
direct
and
indirect
water
combined
.1
Also
provided
are
water
ingestion
amounts
by
water
source.
Sources
include
community
water,
bottled
water,
other
sources,
and
all
sources
combined
(
total
water)
2.
Other
sources
include
water
from
private
household
wells
and
rain
cisterns,
and
household
and
public
springs.
ix
Additionally,
the
report
provides
estimates
of
water
consumption
for
"
all
individuals"
and
for
"
consumers
only".
The
estimates
for
all
individuals
are
based
on
all
survey
respondents
in
the
population
(
or
subpopulation)
under
consideration
including
those
who
reported
no
consumption
of
the
water
from
the
source
under
consideration
during
the
two
survey
days.
The
"
consumers
only"
estimates
are
based
on
only
those
respondents
in
the
population
(
or
subpopulation)
of
interest
who
reported
ingestion
of
the
water
from
the
source
under
consideration
during
the
two
survey
days
and
excludes
the
"
zero"
consumers.
All
estimates
are
provided
in
units
of
milliliters/
person/
day
(
ml/
person/
day)
and
milliliters/
kilogram
of
body
weight/
day
(
ml/
kg/
day).

The
estimated
mean
two
 
day
average
per
capita
ingestion
of
community
water
is
927
ml/
person/
day.

This
mean
ingestion
estimate
applies
to
all
individuals
in
the
United
States
population.
A
90%
confidence
interval
about
this
mean
ingestion
ranges
from
902
to
951
ml/
person/
day
(
See
Table
4
 
1
 
B1).
These
estimates
of
community
water
are
based
on
a
sample
of
15,303
individuals
in
the
50
United
States
and
the
District
of
Columbia.
The
sample
was
selected
to
represent
the
entire
population
of
the
United
States
based
on
1990
census
data.

The
estimated
90th
percentile
of
the
empirical
distribution
of
two
 
day
average
per
capita
ingestion
of
community
water
is
2.016
liters/
person/
day.
The
90%
bootstrap
interval
about
the
90th
percentile
estimate
ranges
from
1.991
to
2.047
liters/
person/
day.
Therefore,
current
ingestion
data
indicate
that
90
percent
of
the
United
States
population
ingests
an
amount
of
community
water
which
is
approximately
less
than
or
equal
to
the
two
liters/
person/
day
estimate
used
as
a
standard
ingestion
value
by
many
federal
agencies
(
See
Table
4
 
1
 
B1).

Women
aged
15
to
44
years,
the
childbearing
years,
ingest
a
mean
of
922
ml
of
community
water
per
day
(
90%
confidence
interval
is
887
to
957
ml).
This
mean
ingestion
is
similar
to
the
mean
daily
per
capita
ingestion
of
community
water
for
the
United
States
population.
Lactating
women
have
the
highest
community
water
ingestion
of
any
subpopulation
identified
in
the
sample.
Lactating
women
reported
a
mean
two
 
day
average
ingestion
of
1.379
liters
(
90%
confidence
interval
is
1.021
to
1.737
ml/
person/
day).

The
90th
and
95th
percentile
estimates
of
ingestion
of
community
water
for
lactating
women
are
2.872
and
3.434
liters/
day,
respectively
(
See
Table
4
 
1
 
E).

The
estimates
of
community
water
ingestion
based
on
"
consumers
only"
are
higher
than
those
based
on
all
individuals
because
respondents
reporting
zero
community
water
ingestion
during
the
two
survey
days
are
excluded
from
the
analysis.
For
"
consumers
only,"
the
estimated
mean
two
 
day
average
per
capita
ingestion
of
community
water
is
1.0
liter/
person/
day
(
90%
confidence
interval
is
976
to
1,024
ml/
person/
day).
These
estimates
are
based
on
the
14,012
respondents
to
the
CSFII
who
reported
consuming
community
water.
The
estimated
90th
percentile
of
consumption
is
2.069
liters/
person/
day
(
See
Table
4
 
2
 
B1).
x
The
highest
consumption
estimates
(
and
therefore
most
conservative
with
regard
to
risk)
are
for
total
water
ingestion
by
"
consumers
only."
The
estimated
mean
ingestion
of
total
water
by
"
consumers
only"
is
1,241
ml/
person/
day
(
90%
confidence
interval
is
1,208
to
1,274
ml/
person/
day).
The
estimated
90th
and
95th
percentiles
are
2,345
ml/
person/
day
and
2,922
ml/
person/
day,
respectively
(
See
Table
4
 
2
 
A).

For
babies
younger
than
one
year
old
the
estimated
mean
community
water
ingestion
is
342
ml/
person/
day
(
90%
confidence
interval
is
295
to
388
ml/
person/
day);
the
estimated
90th
percentile
is
878
ml/
person/
day
(
90%
bootstrap
interval
is
849
to
918
ml/
person/
day);
and
the
95th
percentile
is
1,040
ml/
person/
day
(
90%
bootstrap
interval
is
936
to
1121
ml/
person/
day)
(
See
Table
4
 
1
 
B1).
Thus,
the
standard
one
liter
ingestion
rate
used
in
risk
assessments
for
a
10
 
kilogram
child
is
approximately
less
than
or
equal
to
the
95th
percentile
of
the
empirical
distribution
of
community
water
ingestion
for
infants.

For
babies
younger
than
one
year
old
who
are
water
consumers,
the
estimated
mean
total
water
ingestion
is
563
ml/
person/
day
(
90%
confidence
interval
is
508
to
618
ml/
person/
day).
The
estimated
90th
percentile
is
968
ml/
person/
day
(
90%
bootstrap
interval
is
940
to
1,121
ml/
person/
day),
and
the
estimated
95th
percentile
is
1,236
ml/
person/
day
(
90%
bootstrap
interval
is
1,121
to
1,282
ml/
person/
day).
Thus,
the
one
liter
standard
used
in
risk
assessments
for
a
10
 
kilogram
child
is
approximately
less
than
or
equal
to
the
90th
percentile
of
the
empirical
distribution
of
total
water
ingestion
for
babies
less
than
one
year
old
when
considering
"
consumers
only"
(
See
Table
4
 
2
 
D1).

The
Recommended
Dietary
Allowances
(
RDA,
1989)
for
water
intake
are
1.5
ml/
K
cal
and
980
K
cal/
day
for
a
child
between
six
months
and
one
year
old.
Thus,
the
RDA
for
a
10
 
kilogram
child
is
equivalent
to
1,275
ml
of
water/
day.
Therefore,
the
default
of
1
liter/
10
 
kg
child/
day
is
slightly
lower
than
the
RDA
value
of
1,275
milliliters
per
child
per
day.

For
children
one
to
ten
years
old,
the
estimated
mean
community
water
ingestion
is
400
ml/
person/
day
(
90%
confidence
interval
is
380
to
420
ml/
person/
day);
the
90th
percentile
is
905
ml/
person/
day
(
90%

bootstrap
interval
is
863
to
935
ml/
person/
day)
and
the
95th
percentile
is
1,118
ml/
person/
day
(
90%

bootstrap
interval
is
1,079
to
1,143
ml/
person/
day),
respectively
(
See
Table
4
 
1
 
B1).
Thus,
the
standard
one
liter
ingestion
rate
used
for
risk
assessments
for
a
10
 
kilogram
child
lies
between
the
90th
and
95th
percentiles
of
the
empirical
distribution
of
community
water
ingestion
for
children
one
to
ten
years
old.

For
children
one
to
ten
years
old
who
consume
water,
the
estimated
mean
total
water
ingestion
is
532
ml/
person/
day
(
90%
confidence
interval
is
509
to
556
ml/
person/
day).
The
estimated
90th
percentile
of
total
water
ingestion
is
1,004
ml/
person/
day
(
90%
bootstrap
interval
is
980
to
1,030
ml/
person/
day),
and
the
estimated
95th
percentile
is
1,242
ml/
person/
day
(
90%
bootstrap
interval
is
1,198
to
1,284
ml/
person/
day)
(
See
Table
4
 
2
 
D1).
Thus,
the
one
liter
standard
ingestion
used
in
risk
assessments
for
a
10
 
kilogram
child
is
approximately
less
than
or
equal
to
the
90th
percentile
of
the
empirical
distributions
of
total
water
ingestion
for
children
one
to
ten
years
old
when
considering
"
consumers
only."
xi
When
considering
water
ingestion
rates
based
on
units
of
milliliters
per
kilogram
of
body
weight
per
day,
this
analysis
shows
that
the
mean
ingestion
rates
for
babies
younger
than
one
year
are
estimated
to
be
three
to
four
times
higher
than
the
mean
rates
for
the
population
as
a
whole.
For
example,
the
estimated
community
water
ingestion
rate
is
46
ml/
kg/
day
(
90%
confidence
interval
is
39
to
53
ml/
kg/
day)
for
babies
in
the
U.
S.
population
versus
16
ml/
kg/
day
(
90%
confidence
interval
is
15
to
16
ml/
kg/
day)
for
the
general
population
(
See
Table
4
 
1
 
B2).
The
estimated
community
water
ingestion
rate
for
babies
consuming
community
water
is
69
ml/
kg/
day
(
90%
confidence
interval
is
62
to
77
ml/
kg/
day)
versus
17
ml/
kg/
day
(
90%
confidence
interval
is
16
to
17
ml/
person/
day)
for
the
general
population
(
See
Table
4
 
2
 
B2).

The
mean
per
capita
ingestion
of
community
water
is
75
percent
of
the
mean
total
water
ingested
from
all
sources.
The
mean
bottled
water
ingested
is
13
percent
of
the
mean
of
total
water
ingestion,
while
water
from
other
sources
such
as
wells
and
rain
cisterns
is
10
percent
of
the
mean
of
total
water
ingested.

Many
federal
agencies,
including
EPA,
use
the
standard
water
ingestion
quantities
of
two
liters
for
a
70
 
kilogram
adult
and
one
liter
for
a
10
 
kilogram
child.
This
2
 
liter
quantity
of
ingested
water
is
supported
by
a
National
Cancer
Institute
(
NCI)
analysis
of
the
USDA
1977
 
78
USDA
National
Food
Consumption
Survey
(
NFCS)
data
(
1989,
Ershow
and
Cantor).
The
mean
per
capita
daily
intake
of
tap
water,
as
estimated
from
the
1977
 
78
NFCS
data
is
1.193
liters/
person/
day.
The
estimated
percentile
corresponding
to
two
liters
per
day
ingested
is
the
88th.
There
are
a
number
of
differences
in
the
methodologies
used
in
the
Ershow
and
Cantor
study
and
this
analysis.
One
difference
is
that
the
Ershow
and
Cantor
estimates
were
based
on
1977
 
78
data
while
the
estimates
in
this
document
are
based
on
data
collected
in
1994
through
1996.
A
second
difference
is
that
the
1977
 
78
NFCS
was
based
on
three
consecutive
days
of
food
intake
while
the
1994
 
96
CSFII
was
based
on
two
non
 
consecutive
days.
A
third
difference
is
that
the
Ershow
and
Cantor
report
defined
tap
water
as
"
water
from
the
household
tap."
In
this
report,
water
coming
from
the
tap
is
distinguished
by
source.
Sources
of
water
coming
from
the
tap
may
include:
community
water,
household
well
or
cistern,
a
household
or
public
spring,
and
other.

Another
way
that
the
estimates
in
this
report
differ
from
those
in
the
1989
report
is
that
the
1994
 
96
data
include
water
ingestion
by
pregnant
and
lactating
women.
These
women
were
excluded
from
the
1989
report.

The
CSFII
surveys
have
advantages
and
limitations
for
estimating
per
capita
water
ingestion.
The
primary
advantage
of
the
CSFII
surveys
is
that
they
were
designed
and
conducted
by
the
USDA
to
support
unbiased
estimation
of
food
consumption
across
the
population
in
the
United
States
and
the
District
of
Columbia.
One
limitation
of
the
CSFII
surveys
is
that
individual
food
consumption
data
were
collected
for
only
two
days
 
a
brief
period
which
does
not
necessarily
depict
"
usual
intake."
Usual
dietary
intake
is
defined
as
"
the
long
 
run
average
of
daily
intakes
by
an
individual."
Upper
percentile
estimates
may
differ
for
short
 
term
and
long
 
term
data
because
short
term
food
consumption
data
tend
to
be
inherently
more
variable.
It
is
important
to
note,
however,
that
variability
due
to
duration
of
the
survey
does
not
result
in
bias
of
estimates
of
overall
mean
consumption
levels.
A
second
limitation
is
that
the
multistage
survey
xii
design
does
not
support
interval
estimates
for
many
of
the
subpopulations
reported
in
this
document
because
of
sparse
representation
in
the
sample.
Therefore,
only
mean
and
percentile
estimates
are
reported
for
all
subpopulations
considered
here.
The
survey
does
support
interval
estimates
for
the
U.
S.
population
and
some
large
subpopulations
which
are
presented
in
Chapter
4.
A
third
limitation
is
that
the
survey
design
does
not
support
generating
water
consumption
estimates
for
certain
subpopulations
of
interest.

Examples
of
such
subpopulations
are
Native
Americans
with
traditional
lifestyles,
people
who
live
in
hot
climates,
people
who
consume
large
amounts
of
water
because
of
physical
activity,
and
people
with
medical
conditions
necessitating
increased
water
intake.
While
these
individuals
are
participants
in
the
survey,
they
are
not
present
in
sufficient
numbers
to
support
water
ingestion
estimates.
xiii
THIS
PAGE
INTENTIONALLY
LEFT
BLANK
xiv
TABLE
OF
CONTENTS
Section
Page
ACKNOWLEDGMENTS
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
i
SAB
REVIEWPANELMEMBERS
AND
FEDERAL
EXPERT
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
ii
PREFACE
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
iii
EXECUTIVE
SUMMARY
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
viii
TABLE
OF
CONTENTS
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
xiv
LIST
OF
FIGURES
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
xvi
LIST
OF
KEYTABLES
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
xviii
1.
DEFINITIONS
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
 
1
2.
USDA's
CSFII
SURVEYDESCRIPTION
AND
FILES
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
2
 
1
2.1
SurveyDescription
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
2
 
1
2.2
Dietary
Records
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
2
 
2
2.3
Data
Files
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
2
 
2
2.4
SurveyWeights
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
2
 
3
3.
METHODS
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
3
 
1
3.1
Data
Conventions
Applied
to
the
1994
 
96
CSFII
Data
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
3
 
1
3.2
Statistical
Methods
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
3
 
5
3.3
Rounding
Procedures
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
3
 
7
3.4
Units
of
Measure
Including
Conversion
Factors
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
3
 
7
3.5
Sample
SizeCriteria
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
3
 
7
4.
RESULTS
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
1
4.1
Ingestion
of
Community
Water
a.
ml/
person/
day
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
2
b.
ml/
kg
of
bodyweight/
day
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
2
4.2
Ingestion
of
Community
Water
by
Age
and
Gender
a.
ml/
person/
day
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
3
b.
ml/
kg
of
bodyweight/
day
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
4
4.3
Ingestion
of
Community
Water
for
Children
Younger
Than
One
Year
of
Age
a.
ml/
person/
day
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
4
xv
b.
ml/
kg
of
bodyweight/
day
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
4
4.4
Ingestion
of
Community
Water
for
Women
in
Childbearing
Years
a.
ml/
person/
day
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
4
b.
ml/
kg
of
bodyweight/
day
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
5
4.5
Ingestion
of
BottledWater
andWater
fromOther
Sources
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
6
4.6
Ingestion
of
Total
Water
a.
for
All
Individuals
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
6
b.
for
"
Consumers
Only"
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
7
5.
DISCUSSION
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
5
 
1
5.1
Survey
Strengths
and
Limitations
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
5
 
1
5.2
Sources
of
Error,
Bias,
andUncertainty
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
5
 
3
5.3
Conclusion
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
5
 
4
5.4
References
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
5
 
6
APPENDIX
A:
CSFII
SurveyQuestions
Pertaining
toWater
Ingestion
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
A
 
1
APPENDIX
B:
Examples
of
Procedures
Used
in
the
Estimation
of
IndirectWater
Ingestion
.
.
.
.
.
.
.
B
 
1
APPENDIX
C:
1994
 
96
CSFII
Food
Codes
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
C
 
1
APPENDIX
D:
Statistical
Methods
and
Sample
Design
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
D
 
1
APPENDIX
E:
Per
CapitaWater
Ingestion
Estimates
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
E
 
1
APPENDIX
F:
SAB
Report
and
EPAResponse
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
F
 
1
xvi
LIST
OF
FIGURES
Figure
Page
3­
1.
Water
SourceAssignment
for
DirectWater
Ingestion
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
3
 
9
3­
2.
Data
Conventions
for
Direct
Water
Ingestion
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
3
 
10
3­
3.
Water
SourceAssignment
for
IndirectWater
Ingestion
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
3
 
11
3­
4.
Data
Conventions
for
Indirect
Water
Ingestion
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
3
 
12
4­
1.
All
Individuals
4­
1­
A.
Estimated
Mean
and
90%
Confidence
Intervals
Around
the
Mean
Direct
and
Indirect
Water
Ingestion
­
By
Source
(
ml/
person/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
10
4­
1­
B1.
Estimated
Mean
and
90%
Confidence
Intervals
Around
the
Mean
Direct
and
Indirect
CommunityWater
Ingestion
­
ByAgeCategories
(
ml/
person/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
12
4­
1­
B2.
Estimated
Mean
and
90%
Confidence
Intervals
Around
the
Mean
Direct
and
Indirect
Community
Water
Ingestion
­
By
Age
Categories
(
ml/
kg
of
bodyweight/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
14
4­
1­
C1.
Estimated
Mean
and
90%
Confidence
Intervals
Around
the
Mean
Direct
and
Indirect
CommunityWater
Ingestion
­
ByGender
andAgeCategories
(
ml/
person/
day)
.
.
.
.
.
4
 
16
4­
1­
C2.
Estimated
Mean
and
90%
Confidence
Intervals
Around
the
Mean
Direct
and
Indirect
Community
Water
Ingestion
­
By
Gender
and
Age
Categories
(
ml/
kg
of
bodyweight/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
18
4­
1­
D1.
Estimated
Mean
and
90%
Confidence
Intervals
Around
the
Mean
Total
Direct
and
IndirectWater
Ingestion
­
ByAgeCategories
(
ml/
person/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
20
4­
1­
D2.
Estimated
Mean
and
90%
Confidence
Intervals
Around
the
Mean
Total
Direct
and
IndirectWater
Ingestion
­
ByAgeCategories
(
ml/
kg
of
bodyweight/
day)
.
.
.
.
.
.
.
.
.
4
 
22
4­
1­
E.
Estimated
Mean
and
90%
Confidence
Intervals
Around
the
Mean
Direct
and
Indirect
Community
Water
Ingestion
­
By
Pregnant,
Lactating
and
Childbearing
Age
Women
Categories
(
ml/
person/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
24
4­
1­
F1.
Cumulative
Distribution
of
Per
Capita
Direct
and
Indirect
Community
Water
Ingestion
(
ml/
person/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
25
4­
1­
F2.
Cumulative
Distribution
of
Per
Capita
Direct
and
Indirect
Community
Water
Ingestion
(
ml/
kg
of
bodyweight/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
26
4­
1­
G1.
Histogram
of
Per
Capita
Direct
and
Indirect
Community
Water
Ingestion
(
ml/
person/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
27
4­
1­
G2.
Histogram
of
Per
Capita
Direct
and
Indirect
Community
Water
Ingestion
(
ml/
kg
of
bodyweight/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
28
4­
2.
Consumers
Only
4­
2­
A.
Estimated
Mean
and
90%
Confidence
Intervals
Around
the
Mean
Direct
and
Indirect
Water
Ingestion
­
By
Source
(
ml/
person/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
30
xvii
4­
2­
B1.
Estimated
Mean
and
90%
Confidence
Intervals
Around
the
Mean
Direct
and
Indirect
CommunityWater
Ingestion
­
ByAgeCategories
(
ml/
person/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
32
4­
2­
B2.
Estimated
Mean
and
90%
Confidence
Intervals
Around
the
Mean
Direct
and
Indirect
Community
Water
Ingestion
­
By
Age
Categories
(
ml/
kg
of
bodyweight/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
34
4­
2­
C1.
Estimated
Mean
and
90%
Confidence
Intervals
Around
the
Mean
Direct
and
Indirect
CommunityWater
Ingestion
­
ByGender
andAgeCategories
(
ml/
person/
day)
.
.
.
.
.
4
 
36
4­
2­
C2.
Estimated
Mean
and
90%
Confidence
Intervals
Around
the
Mean
Direct
and
Indirect
Community
Water
Ingestion
­
By
Gender
and
Age
Categories
(
ml/
kg
of
bodyweight/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
38
4­
2­
D1.
Estimated
Mean
and
90%
Confidence
Intervals
Around
the
Mean
Total
Direct
and
IndirectWater
Ingestion
­
ByAgeCategories
(
ml/
person/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
40
4­
2­
D2.
Estimated
Mean
and
90%
Confidence
Intervals
Around
the
Mean
Total
Direct
and
IndirectWater
Ingestion
­
ByAgeCategories
(
ml/
kg
of
bodyweight/
day)
.
.
.
.
.
.
.
.
.
4
 
42
4­
2­
E.
Estimated
Mean
and
90%
Confidence
Intervals
Around
the
Mean
Direct
and
Indirect
Community
Water
Ingestion
­
By
Pregnant,
Lactating
and
Childbearing
Age
Women
Categories
(
ml/
person/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
44
4­
2­
F1.
Cumulative
Distribution
of
Per
Capita
Direct
and
Indirect
Community
Water
Ingestion
(
ml/
person/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
45
4­
2­
F2.
Cumulative
Distribution
of
Per
Capita
Direct
and
Indirect
Community
Water
Ingestion
(
ml/
kg
of
bodyweight/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
46
4­
2­
G1.
Histogram
of
Per
Capita
Direct
and
Indirect
Community
Water
Ingestion
(
ml/
person/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
47
4­
2­
G2.
Histogram
of
Per
Capita
Direct
and
Indirect
Community
Water
Ingestion
(
ml/
kg
of
bodyweight/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
48
xviii
LIST
OF
KEY
TABLES
Table
Page
4­
1.
All
Individuals
4­
1­
A.
EstimatedDirect
and
IndirectWater
Ingestion
­
By
Source
(
ml/
person/
day)
.
.
.
.
.
.
.
.
4
 
9
4­
1­
B1.
Estimated
Direct
and
Indirect
Community
Water
Ingestion
­
By
Age
Categories
(
ml/
person/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
11
4­
1­
B2.
Estimated
Direct
and
Indirect
Community
Water
Ingestion
­
By
Age
Categories
(
ml/
kg
of
bodyweight/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
13
4­
1­
C1.
Estimated
Direct
and
Indirect
Community
Water
Ingestion
­
By
Gender
and
AgeCategories
(
ml/
person/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
15
4­
1­
C2.
Estimated
Direct
and
Indirect
Community
Water
Ingestion
­
By
Gender
and
AgeCategories
(
ml/
kg
of
bodyweight/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
17
4­
1­
D1.
Estimated
Total
Direct
and
Indirect
Water
Ingestion
­
By
Age
Categories
(
ml/
person/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
19
4­
1­
D2.
Estimated
Total
Direct
and
Indirect
Water
Ingestion
­
By
Age
Categories
(
ml/
kg
of
bodyweight/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
21
4­
1­
E.
Estimated
Direct
and
Indirect
Community
Water
Ingestion
­
By
Pregnant,

Lactating
and
ChildbearingAgeWomen
Categories
(
ml/
person/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
23
4­
2.
Consumers
Only
4­
2­
A.
EstimatedDirect
and
IndirectWater
Ingestion
­
By
Source
(
ml/
person/
day)
.
.
.
.
.
.
.
4
 
29
4­
2­
B1.
Estimated
Direct
and
Indirect
Community
Water
Ingestion
­
By
Age
Categories
(
ml/
person/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
31
4­
2­
B2.
Estimated
Direct
and
Indirect
Community
Water
Ingestion
­
By
Age
Categories
(
ml/
kg
of
bodyweight/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
33
4­
1­
C1.
Estimated
Direct
and
Indirect
Community
Water
Ingestion
­
By
Gender
and
AgeCategories
(
ml/
person/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
35
4­
1­
C2.
Estimated
Direct
and
Indirect
Community
Water
Ingestion
­
By
Gender
and
AgeCategories
(
ml/
kg
of
bodyweight/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
37
4­
2­
D1.
Estimated
Total
Direct
and
Indirect
Water
Ingestion
­
By
Age
Categories
(
ml/
person/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
39
4­
2­
D2.
Estimated
Total
Direct
and
Indirect
Water
Ingestion
­
By
Age
Categories
(
ml/
kg
of
bodyweight/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
41
4­
2­
E.
Estimated
Direct
and
Indirect
Community
Water
Ingestion
­
By
Pregnant,

Lactating
and
ChildbearingAgeWomen
Categories
(
ml/
person/
day)
.
.
.
.
.
.
.
.
.
.
.
.
.
4
 
43
xix
THIS
PAGE
INTENTIONALLY
LEFT
BLANK
1­
1
1.
DEFINITIONS
BiologicalWater
is
water
found
naturally
in
foods.
This
water
source
is
not
included
in
the
estimates
presented
in
this
document.

Bottled
Water
is
purchased
plain
water.

Broad
Age
Categories
cover
babies
(
less
than
one
year
old),
children
(
one
to
10
years
old),
young
adults
(
11
to
19
years
old),
adults
(
20
years
and
older).

CommercialWater
is
water
added
by
the
manufacturer
prior
to
merchandising.
This
water
is
not
included
in
the
estimates
presented
in
this
document.
An
example
of
commercial
water
is
water
added
to
bottled
iced
tea
by
the
manufacturer.

Community
Water
is
tap
water
from
the
community
water
supply.

Consumers
are
individuals
who
reported
ingestion
of
the
water
source
under
consideration.
Individuals
with
reported
ingestions
of
zero
are
not
considered
consumers.

CSFII
is
the
acronym
for
the
USDA's
Continuing
Survey
of
Food
Intakes
by
Individuals.

Direct
Water
is
plain
water
ingested
directly
as
a
beverage.

Fine
Age
Categories
include
11
age
groupings.
These
groupings
are
less
than
six
months
(<
0.5
years),

between
six
months
and
one
year
(
0.5
to
0.9
years),
1
to
3
years,
4
to
6
years,
7
to
10
years,
11
to
14
years,
15
to
19
years,
20
to
24
years,
25
to
54
years,
55
to
64
years,
and
65
and
older.

Food
Code
is
an
8
 
digit
number
assigned
to
each
unique
food
in
the
USDA
Food
Coding
Database.

Food
Coding
Database
is
a
database
in
the
USDA
CSFII
1994
 
96
Technical
Support
Files
which
contains
information
used
to
code
foods
and
amounts,
including
descriptions
of
the
food
code.

Indirect
Water
is
water
added
to
foods
and
beverages
during
final
preparation
at
home,
or
by
food
service
establishments
such
as
school
cafeterias
and
restaurants.
An
example
of
indirect
water
is
water
added
to
dry
cake
mix.
1­
2
Missing
Water
Source
indicates
that
a
survey
participant
responded
"
don't
know"
or
"
not
ascertained"
to
the
survey
question
regarding
the
source
of
water.

Nutrient
Database
is
a
database
in
the
USDA
1994
 
96
CSFII
Technical
Support
Files
which
contains
nutrient
composition
information
(
including
grams
of
water
per
100
grams
of
food)
used
to
calculate
the
nutrient
value
of
foods
ingested
in
the
CSFII.

Other
Water
is
water
obtained
from
one
of
the
following
sources:
a
well
or
rain
cistern
(
household's),

spring
(
household's
or
public),
or
other
source.

PreparationWater
is
water
used
to
prepare
foods.
Preparation
water
includes
the
water
used
to
prepare
foods
at
home
and
by
local
food
service
establishments
(
indirect
water),
as
well
as,
water
added
by
commercial
food
manufacturers.

Recipe
Database
is
a
database
in
the
USDA
1994
 
96
CSFII
Technical
Support
Files
which
provides,
for
each
food
code,
a
standard
recipe
including
the
gram
weight
of
each
ingredient.

Technical
Support
Files
consist
of
four
USDA
technical
databases
used
to
code
food
data
collected
in
the
1994
 
96
CSFII.
They
are
the
Food
Coding
Database,
the
Nutrient
Database,
the
Recipe
Database,
and
the
Pyramid
Servings
Database.

TotalWater
is
the
sum
of
direct
and
indirect
water
from
all
sources.
Water
sources
include
community
water,
bottled
water,
other
water
and
missing
sources.



2.
USDA's
CSFII
SURVEY
DESCRIPTION
AND
FILES
A
brief
description
of
the
USDA's
1994,
1995,
and
1996
series
of
the
Continuing
Survey
of
Food
Intakes
by
Individuals
(
CSFII)
is
presented
in
this
chapter.
Section
2.1
provides
a
description
of
the
surveys,
and
Section
2.2
describes
the
process
used
to
collect
the
dietary
recall
information.
Files
from
which
data
were
drawn
to
produce
the
estimates
in
this
report
are
listed
in
Section
2.3.
Section
2.4
presents
a
brief
discussion
about
survey
weights
and
their
use.
The
details
of
the
sample
design
and
resulting
survey
weights
are
provided
in
Appendix
D.

2.1
Survey
Description
The
CSFII,
conducted
by
the
United
States
Department
of
Agriculture
(
USDA),
collects
dietary
intake
information
from
nationally
representative
samples
of
non
 
institutionalized
persons
residing
in
United
States
households.
Households
in
these
national
surveys
are
sampled
from
the
50
states
and
Washington,

D.
C.
Each
survey
collects
daily
consumption
records
for
approximately
10,000
food
codes
across
nine
food
groups.
These
food
groups
are
(
1)
milk
and
milk
products;
(
2)
meat,
poultry,
and
fish;
(
3)
eggs;
(
4)

dry
beans,
peas,
legumes,
nuts,
and
seeds;
(
5)
grain
products;
(
6)
fruit;
(
7)
vegetables;
(
8)
fats,
oils,
and
salad
dressings;
and
(
9)
sweets,
sugars,
and
beverages.
Data
provide
"
up
 
to
 
date
information
on
food
intakes
by
Americans
for
use
in
policy
formation,
regulation,
program
planning
and
evaluation,
education,

and
research."
The
survey
is
"
the
cornerstone
of
the
National
Nutritional
Monitoring
and
Related
Research
Program,
a
set
of
related
federal
activities
intended
to
provide
regular
information
on
the
nutritional
status
of
the
United
States
population"
(
CSFII
survey
documentation,
p.
2
 
3).

The
1994
 
1996
CSFII
was
conducted
according
to
a
stratified,
multi
 
area
probability
sample
organized
using
estimates
of
the
1990
United
States
population.
Stratification
accounted
for
geographic
location,
degree
of
urbanization,
and
socioeconomics.
Each
year
of
the
survey
consisted
of
one
sample
with
oversampling
for
low
 
income
households.
Eligibility
for
the
low
 
income
sample
was
limited
to
households
with
gross
incomes
at
or
below
130
percent
of
the
federal
poverty
guidelines
(
DHHS
1996).

The
sample
design
aimed
at
specified
precision
levels
for
estimates
of
mean
one
 
day
consumption
of
saturated
fat
and
iron.

Two
days
of
dietary
recall
data
were
provided
by
15,303
individuals
across
the
three
survey
years.

This
constitutes
an
overall
two
 
day
response
rate
of
75.9
percent.
Response
rates
for
each
survey
year
are
provided
in
Table
2
 
1.
Survey
weights
were
corrected
by
the
USDA
for
nonresponse.



TABLE
2
 
1
CSFII
RESPONSE
RATES
YEAR
TOTAL
ELIGIBLE
INDIVIDUALS
SAMPLED
NUMBER
WITH
TWO
 
DAY
RESPONSE
(
TWO
 
DAY)
RESPONSE
RATE
1994
6,973
5,311
76.2%

1995
6,664
5,072
76.1%

1996
6,484
4,920
75.9%

2.2
Dietary
Records
Survey
participants
provided
two
non
 
consecutive,
24
 
hour
days
of
dietary
data.
Both
days'
dietary
recall
information
was
collected
by
an
in
 
home
interviewer.
Interviewers
provided
participants
with
an
instructional
booklet
and
standard
measuring
cups
and
spoons
to
assist
them
in
adequately
describing
the
type
and
amount
of
food
ingested.
If
the
respondent
referred
to
a
cup
or
bowl
in
their
own
home,
a
2
 
cup
measuring
cup
was
provided
to
aid
in
the
calculation
of
the
amount
consumed.
The
sample
person
could
fill
their
own
bowl
or
cup
with
water
to
represent
the
amount
eaten
or
drunk,
and
the
interviewer
could
then
measure
the
amount
consumed
by
pouring
it
into
the
2
 
cup
measure.
The
Day
2
interview
occurred
3
to
10
days
after
the
Day
1
interview,
but
not
on
the
same
day
of
the
week.
The
interviews
allowed
participants
"
three
passes"
through
the
daily
intake
record
to
maximize
recall
(
CSFII
survey
documentation,
p.
3
 
6).
Proxy
interviews
were
conducted
for
children
aged
six
and
younger
and
sampled
individuals
unable
to
report
due
to
mental
or
physical
limitations.
The
average
questionnaire
administration
time
for
Day
1
intake
was
30
minutes,
while
Day
2
averaged
27
minutes.

2.3
Data
Files
The
USDA
records
1994
 
96
CSFII
participant
information
in
three
record
types.
Data
extracted
from
these
record
types
provide
the
information
to
determine
the
amount
and
source
of
commodities
ingested
by
participants.
These
data
are
publicly
available
on
CD
 
ROM
(
See
Section
5.4
References),
and
the
three
CSFII
record
types
used
for
this
report
are
described
here.
Record
type
15
(
RT15)
reports
household
information.
Generally
the
source
of
water
is
determined
from
these
records.
Record
type
25
(
RT25)

records
individual
information.
This
is
where
the
amount
of
direct
water
ingested
is
recorded.
Record
type
30
(
RT30)
records
food
items
ingested
on
each
of
the
two
survey
days
by
each
individual.
The
amount
of
indirect
water
ingested
can
be
calculated
from
these
records
in
conjunction
with
the
CSFII
1994
 
96
Technical
Support
Files
including
the
food
coding,
recipe
and
nutrient
databases.
Refer
to
Appendix
A
for
the
CSFII
questions
related
to
the
amount
of
water
ingested
and
the
source
of
the
water.
Chapter
3
details
how
these
record
types
were
combined
to
establish
a
working
database
of
individual
records
with
the
amount,
source,
and
type
of
ingestion
(
direct
or
indirect).



2.4
Survey
Weights
USDA
files
provide
a
survey
weight
for
each
individual
with
two
days
of
consumption
data
in
the
1994­
96
survey.
These
weights
account
for
the
probability
that
the
individual
was
selected
and
contain
adjustments
for
non­
respondents.
The
recorded
weights
also
reflect
USDA's
calibration
to
ensure
that
the
sample
is
representative
of
population
characteristics
during
the
three
years
of
the
survey.
Survey
weights
are
applied
during
the
generation
of
ingestion
estimates
recorded
in
this
report.
These
weights
project
data
from
an
individual
to
the
population.
Appendix
D
provides
a
more
detailed
discussion
of
the
development
and
application
of
the
three
year,
two
day
surveyweights.



THIS
PAGE
INTENTIONALLY
LEFT
BLANK
3­
1
3.
METHODS
This
chapter
addresses
the
methods
employed
to
produce
the
reported
daily
average
per
capita
water
ingestion
estimates
by
source
and
type
of
ingestion.
Section
3.1
defines
the
data
conventions
applied
to
the
CSFII
data
to
establish
water
ingestion
records
for
each
respondent
with
two
days
of
consumption
data.

These
conventions
include
identification
of
the
records
from
CSFII
used
to
determine
the
source
and
amount
of
directly
ingested
waters.
We
also
describe
CSFII
auxiliary
files
and
how
they
were
used
to
quantify
the
amount
of
indirect
water
ingested
by
a
survey
participant.
Section
3.2
summarizes
the
statistical
methods
used
to
estimate
the
mean
and
percentiles
of
the
empirical
distributions
of
daily
average
per
capita
water
ingestion.
Appendix
D
provides
the
statistical
formulae
used
to
calculate
these
estimates.

Rounding
procedures
and
units
of
measurement
are
recorded
in
Sections
3.3
and
3.4,
respectively.
Section
3.5
provides
the
minimum
sample
size
requirements
used
to
identify
potentially
unreliable
estimates.

3.1
Data
Conventions
Applied
to
the
1994
 
96
CSFII
Data
A
series
of
CSFII
records
is
used
to
define
the
source
and
type
of
water
ingested
by
a
survey
respondent.
We
drew
from
household
records
and
individual
records
to
define
the
amount
and
source
of
plain
water
ingested
as
a
beverage.
The
source
and
amount
of
indirect
water
ingestion
was
determined
using
the
household
and
individual
records
in
conjunction
with
the
1994
 
96
CSFII
Technical
Support
Files
including
the
food
coding,
recipe
and
nutrient
databases.
All
CSFII
data
used
are
publicly
available
on
CD
 
ROM
(
See
Section
5.4
References).
The
following
paragraphs
describe
the
protocols
followed
for
assigning
the
source
and
quantifying
the
amount
of
the
daily
average
water
ingested
by
each
respondent
with
two
days
of
consumption
records.

CSFII
record
type
15
(
RT15)
reports
household
information.
The
source
of
water
ingested
is
generally
assigned
from
these
records.
Record
type
25
(
RT25)
records
the
amount
of
direct
water
ingested.
Record
type
30
(
RT30)
reports
food
items
ingested
on
each
of
the
two
survey
days
by
each
individual.
The
amount
of
indirect
water
ingested
by
each
participant
was
calculated
from
food
code
records
in
conjunction
with
the
1994
 
96
CSFII
Technical
Support
Files.
A
more
detailed
description
of
how
information
was
drawn
from
these
three
sets
of
records
to
determine
the
average
daily
water
ingestion
for
each
survey
participant
is
provided
in
the
remainder
of
this
chapter.
We
first
describe
the
data
conventions
and
then
follow
the
description
with
a
flow
chart.
As
a
point
of
reference,
Appendix
A
contains
all
CSFII
questions
related
to
the
amount
of
water
ingested
and
the
source
of
the
water.

Water
Ingestion
listed
as
"
Direct"
is
defined
as
plain
water
directly
ingested
by
an
individual.
The
amount
of
water
ingested
is
recorded
in
CSFII
RT25,
variables
D1_
H2O_
O
and
D2_
H2O_
O.
The
number
following
the
letter
D
in
these
variables
indicates
the
day
of
the
survey
to
which
the
consumption
corresponds.
It
is
in
these
two
variables
that
the
amount
of
direct
water
ingested
by
participants
is
recorded
3­
2
in
fluid
ounces
(
fl
oz).
This
amount
was
converted
to
milliliters
by
multiplying
the
amount
in
fluid
ounces
by
the
conversion
factor
of
29.574
ml/
fl
oz.

The
companion
RT25
variables
to
which
respondents
to
the
question
(
D1_
H2O_
O)
"
How
many
fluid
ounces
of
plain
drinking
water
did
you
consume?"
are
directed
are
D1_
H2O_
H
and
D1_
H2O_
A.
A
similar
set
of
variables
records
information
for
D2_
H2O_
O.
The
variable
with
the
suffix
"
H"
asks
the
respondent
how
much
of
this
water
was
ingested
at
home.
The
choice
of
responses
is
all,
most,
some,
none,

don't
know,
and
not
ascertained.
The
variable
D1_
H2O_
A
asks
what
the
source
was
of
plain
drinking
water
that
did
not
come
from
your
home.
The
choices
of
responses
are
tap/
fountain,
bottled,
other,
don't
know,
and
not
ascertained.
If
an
individual
answers
with
either
of
the
last
two
responses,
the
source
of
that
water
is
considered
"
missing."

Because
the
amount
of
plain,
noncarbonated
water
ingested
by
an
individual
as
recorded
in
the
RT25
files
does
not
completely
designate
the
source
of
the
water,
RT15
household
records
were
consulted.
The
RT15
variable,
H2O_
DRNK,
records
source
information
for
the
household.
For
this
variable,
the
following
conventions
were
applied
to
assign
source.

If
H2O_
DRNK
is
valued
as
1,
then
the
water
source
was
a
community
water
supply
2,
then
the
water
source
was
a
household
well
or
rain
cistern
3,
then
the
water
source
was
a
household
or
public
spring
4,
then
the
water
source
was
considered
bottled
water
(
purchased)

96,
it
is
defined
explicitly
as
"
other"
and
considered
to
be
"
other"
water
sources.

All
remaining
values
of
the
associated
variable,
which
include
98
for
"
don't
know"
and
99
for
"
not
ascertained,"
are
considered
missing
water
sources.

To
determine
source
for
direct
water
ingestion
(
D1_
H2O_
O),
if
RT25
variable
D1_
H2O_
H
is
valued
as
"
1"
designating
"
all,"
then
the
source
was
derived
from
RT15
variable
H2O_
DRNK.

"
2"
designating
"
most,"
then
75%
of
the
water
ingested
was
allocated
according
to
the
RT15
variable
H2O_
DRNK
and
25%
according
to
the
response
to
RT25
variable
D1_
H2O_
A.

"
3"
designating
"
some,"
then
25%
was
allocated
according
to
the
RT15
variable
H2O_
DRNK
and
75%

according
to
the
response
to
RT25
variable
D1_
H2O_
A.

"
4"
designating
"
none,"
then
the
source
was
derived
from
RT25
variable
D1_
H2O_
A.

"
8"
or
"
9"
designating
"
don't
know"
or
"
not
ascertained,"
respectively,
then
50%
was
allocated
according
to
the
RT15
variable
H2O_
DRNK
and
50%
according
to
the
response
to
RT25
variable
D1_
H2O_
A.
3­
3
Indirect
water
is
defined
as
water
added
to
foods
and
beverages
during
final
preparation
at
home
or
by
local
food
service
establishments
(
e.
g.,
school
cafeterias
and
restaurants).
Excluded
from
indirect
water
are
biological
water
and
water
added
by
the
manufacturer
during
processing.
For
example,
an
apple
contains
biological
water,
and
canned
ready
B
to
B
serve
soup
contains
water
added
by
the
manufacturer.
The
1994
B
96
CSFII
Food
Coding
Database
contains
10,620
food
codes.
The
food
code
descriptions
contained
in
USDA's
Food
Coding
Database
generally
do
not
indicate
where
the
food
was
prepared.
Therefore,
in
order
to
identify
indirect
water
ingestion,
each
food
code
description,
corresponding
recipe
and
in
some
instances
nutrient
composition
information
associated
with
the
reported
food
codes
for
the
1994
 
96
CSFII
was
reviewed.
A
subset
of
these
food
codes
which
contained
preparation
water
was
created.
A
food
code
was
considered
to
contain
preparation
water
if
the
food
code
recipe
contained
one
of
the
following
ingredients:
(
1)
water;
(
2)
an
ingredient
which
had
its
own
recipe
which
contained
water;
(
3)
brewed
coffee
or
tea;
and
(
4)
pre
 
cooked
pasta,
rice,
cereals,
beans
or
legumes.
The
subset
consisted
of
7,560
food
codes
which
contained
preparation
water.
The
food
codes
in
this
subset
were
then
reviewed
to
identify
and
exclude
those
which
appeared
to
be
commercial
products
(
e.
g.,
yogurt,
frozen
milk
desserts,
frozen
entrees,

ready
B
to
B
serve
soups,
ready
B
to
B
serve
fruitades
and
drinks,
all
soft
drinks,
and
other
food
codes
with
descriptions
identifying
brand
names).
This
resulted
in
a
smaller
subset
of
2,478
food
codes
which
were
assumed
to
contain
indirect
water.
Next
the
foods
which
could
reasonably
be
assumed
to
have
been
prepared
in
final
form
in
the
home
or
by
a
food
service
establishment
were
identified
(
e.
g.,
foods
described
as
"
made
from
home
recipe,"
orange
juice
made
from
concentrate,
infant
formula
made
from
concentrate,

canned
soup
with
water
added).
It
was
assumed
that
the
recipe
water
in
such
foods
was
100%
indirect.

For
some
foods,
both
homemade
and
commercially
prepared
varieties
were
identified
under
one
food
code.

For
these
food
codes,
a
"
best
guess"
estimate
was
made
as
to
the
proportion
which
would
have
been
home
B
prepared
versus
commercially
processed.
For
example,
it
was
estimated
that
50%
of
pre
B
cooked
beans
to
be
home
B
prepared
and
50%
to
be
commercially
canned.
These
allocations
are
documented
in
Appendix
C1.

When
a
respondent
supplied
specific
information
about
ingredients
that
differed
from
the
standard
recipe
maintained
in
the
Food
Coding
Database,
this
modification
was
recorded.
This
flexibility
allowed
the
database
to
capture
the
specific
type
of
fat,
type
of
milk,
and
dilution
of
foods.
For
example,
if
the
standard
recipe
in
the
Food
Coding
Database
for
an
infant
formula
prepared
from
liquid
concentrate
calls
for
a
specified
amount
of
water
to
be
added
and
a
respondent
reported
making
the
formula
with
3
times
that
amount
of
water,
a
recipe
modification
would
be
created
to
allow
for
this
deviation
from
the
standard
recipe.

Appendix
B1
contains
examples
for
estimating
the
proportion
of
indirect
water
in
100
grams
of
a
food.

The
ingredient
amount
as
a
percent
of
the
prepared
product
(
P%)
was
calculated
for
each
ingredient
of
each
recipe
that
contains
indirect
water
using
the
method
provided
in
USDA
guidance
examples.
Appendix
B3
contains
these
guidance
documents.
The
grams
absorbed
moisture
per
100
grams
cooked
ingredient
3­
4
(
G_
am)
was
calculated
for
pre
B
cooked
pasta,
rice,
cereals,
beans,
and
legumes
using
the
total
solids
method
provided
by
the
USDA
(
refer
to
Appendix
B2).

Next,
the
proportion
of
moisture
in
100
grams
of
food
as
ingested
(
P_
m)
was
found.
These
values
were
taken
from
a
file
(
WTR_
FC.
TXT)
provided
by
the
USDA
when
available.
TheWTR_
FC.
TXT
file
contains
the
amounts
of
water
in
100
grams
of
the
CSFII
1994
B
96
foods.
These
amounts
represent
both
water
from
survey
recipes
as
well
as
from
ingredients
(
referred
to
as
PDS
ingredients)
used
in
the
survey
recipes.
Adjustments
were
made
by
USDA
for
any
moisture
and
fat
losses/
gains
associated
with
the
recipe
in
which
the
PDS
codes
with
water
appear.
For
those
recipe
ingredients
not
available
in
WTR_
FC.
TXT,

the
values
were
calculated
as
follows:

P_
m
=
(
P%)(
G_
am/
100),
for
pre
B
cooked
pasta,
rice,
cereals,
beans,
and
legumes
P_
m
=
P%,
for
water,
brewed
coffee
and
tea,
and
pds
B
coded
ingredients
Then
the
proportion
of
indirect
preparation
water
per
100
grams
of
food
(
G_
i)
was
calculated
for
each
ingredient.
This
was
done
by
multiplying
the
proportion
of
moisture
in
100
grams
of
a
food
as
ingested
(
P_
m)
by
the
percentage
of
that
ingredient
assumed
to
be
home
or
food
service
establishment
prepared
and
dividing
by
100.
(
Appendix
B1
provides
examples
of
these
calculations.)

For
recipes
with
indirect
water,
the
ratio
of
the
amount
of
water
to
the
total
grams
in
the
recipe
was
derived
by
summing
the
values
of
G_
i
across
all
ingredients
in
the
recipe.
This
water
ratio
was
then
multiplied
by
the
amount
of
the
given
food
ingested
by
the
respondent
to
determine
the
number
of
grams
of
indirect
water.
Under
the
assumption
that
the
density
of
this
water
is
1,
the
number
of
grams
of
indirect
water
ingested
from
foods
or
beverages
was
converted
to
milliliters.

To
assign
the
source
of
indirect
water,
several
variables
were
consulted.
First,
if
the
respondent
indicated
in
RT30
variable
FOODSRCE
that
the
source
was
>
1,
then
the
source
was
assumed
to
be
tap
water.
If
FOODSRCE=
1
to
indicate
that
the
food
items
were
obtained
from
the
store,
then
it
was
assumed
the
recipe
was
prepared
at
home.
In
this
case,
RT15
variables
H2O_
COOK
or
H2O_
BEVR
were
consulted.
If
the
first
three
digits
of
the
food
code
indicated
that
the
ingested
food
was
a
beverage,
then
the
water
source
was
assigned
to
the
record
based
on
the
response
to
H2O_
BEVR.
This
question
indicated,

"
What
is
the
main
source
of
the
water
used
in
your
home
for
preparing
beverages
such
as
coffee,
tea,

juices,
and
baby
formula?"
The
same
source
allocations
in
RT15
variable
H2O_
DRNK
were
applied
to
these
records.
Likewise,
if
the
first
three
digits
of
the
food
code
indicated
that
the
food
code
was
not
a
beverage,
then
the
source
was
assigned
according
to
the
response
to
H2O_
COOK,
which
asked,
"
What
is
the
main
source
of
the
water
used
for
cooking
in
your
home?"

Figures
3
B
1
through
3
B
4
present
flow
charts
of
the
data
conventions
for
the
assignment
of
water
source.
3­
5
For
each
of
the
15,303
respondents
with
two
days
of
records
in
the
CSFII
databases,
a
daily
average
ingestion
value
was
determined
for
each
water
source
and
ingestion
type
(
direct,
indirect,
and
both
direct
and
indirect).
For
subpopulation
estimates,
if
a
respondent
was
a
member
of
the
subpopulation
but
did
not
report
ingestion
of
the
specified
water
source
and
ingestion
type,
then
that
individual's
average
daily
amount
of
water
ingested
entered
the
estimation
algorithms
as
zero.
These
estimates
are
provided
in
the
tables
of
this
report
identified
as
"
All
Individuals."

Ingestion
(
direct,
indirect,
and
both
direct
and
indirect)
was
also
estimated
for
consumers
with
two
days
of
records
in
the
CSFII
databases.
Hence,
these
estimates
do
not
include
individuals
who
reported
zero
amounts
of
water
ingested
from
the
water
source
under
consideration.
These
estimates
are
provided
in
the
tables
of
this
report
identified
as
"
Consumer
Only."

The
convention
described
in
the
preceding
paragraphs
produces
individual
daily
averages
in
milliliters/
person/
day.
If
estimates
are
required
on
the
milliliters/
kilogram
body
weight/
day
basis,
then
the
individual's
daily
average
is
divided
by
the
individual's
body
weight
in
kilograms.
The
milliliters/
kilogram
body
weight
daily
average
for
each
individual
then
enters
the
estimating
algorithm
described
in
Section
3.2
and
Appendix
D,
as
do
the
milliliter
daily
averages.

Internal
quality
assurance
and
quality
control
procedures
were
utilized
during
the
calculation
of
estimates
for
this
report.
Algorithm
testing
was
conducted
for
data
procedures.
Data
subsetting
procedures
were
quality
assessed
by
intermediate
estimates
verification.
Final
tabulated
estimates
were
reviewed
for
consistency
and
validity.
USDA
experts
were
consulted
on
data
assumptions.

3.2
Statistical
Methods
This
section
summarizes
the
statistical
methods
used
to
generate
point
and
interval
estimates
of
daily
average
per
capita
water
ingestion.
Point
estimates
include
the
mean,
1st,
5th,
10th,
25th,
50th,
75th,
90th,

95th,
and
99th
percentiles.
Mean
estimates
were
generated
using
ratio
estimation
techniques.
The
mean,

daily
average
per
capita
ingestion
for
a
given
commodity
type,
was
estimated
as
the
ratio
of
total
ingestion
by
the
United
States
population
or
subpopulation,
divided
by
the
estimate
of
the
total
number
of
individuals
in
the
population
or
subpopulation.
Empirical
percentiles
were
estimated
using
nonparametric
techniques.

All
estimates
incorporated
CSFII
survey
weights
to
project
a
sampled
individual's
ingestion
to
the
population.

The
majority
of
the
per
capita
water
ingestion
estimates
in
this
report
are
presented
for
specific
subpopulations
and
water
source.
The
design
of
the
CSFII
survey
did
not
always
support
estimation
of
the
variance
when
subpopulations
were
evaluated.
Without
a
variance
estimate,
confidence
intervals
about
the
mean
or
bootstrap
intervals
about
percentile
estimates
cannot
be
produced.
Therefore,
the
tabulated
presentations
in
Appendix
E
include
only
point
estimates.
However,
the
survey
did
support
variance,
and
3­
6
thus
interval
estimation,
for
some
subpopulations.
These
estimates
are
presented
in
the
key
figures
of
Chapter
4
augmenting
tabulated
estimates
for
the
all
individuals.

When
a
variance
was
estimated
for
the
mean
per
capita
ingestion,
we
estimated
the
variance
of
the
mean
using
a
Taylor
series
approximation
of
the
deviation
of
estimates
from
their
expected
values.
The
Taylor
series
approximations
were
applied
to
ultimate
clusters,
which
resulted
in
an
overall
estimate
of
the
variance
instead
of
estimating
variance
components
due
to
sample
B
design
stages.
In
Appendix
D,
we
include
the
statistical
formulae
for
generating
both
the
mean
estimate
and
the
estimate
of
the
confidence
interval
about
the
mean.
We
also
provide
the
method
for
generating
percentile
estimates
and
estimates
of
90%
bootstrap
intervals
about
the
percentile
estimates.

All
three
CSFII
surveys
are
multistage,
stratified
B
cluster
samples.
Sample
weights,
which
project
the
data
from
a
sampled
individual
to
the
population,
are
based
on
the
probability
of
an
individual
being
sampled
at
each
stage
of
the
sampling
design.
As
mentioned
in
Chapter
2
of
this
report,
the
sample
weights
associated
with
each
individual
reporting
two
days
of
consumption
data
were
adjusted
to
correct
for
nonresponse
bias.
These
adjusted
sample
weights,
which
are
recorded
in
the
CSFII
data
in
the
variable
SAM_
WT,
record
the
number
of
individuals
the
sampled
person
represents
in
the
population.
For
example,
a
sample
weight
valued
as
22
projects
the
data
from
the
individual
with
that
sample
weight
to
22
individuals
in
the
population
of
the
50
United
States
and
the
District
of
Columbia.

Because
the
sample
design
contains
multiple
levels,
specific
information
is
necessary
to
partition
the
variance
B
of
B
the
B
mean
estimate
into
components.
That
is,
specification
of
the
sample
size
and
population
size
within
each
level
of
sampling
is
required.
However,
this
information
is
not
inherent
in
the
CSFII
data.

Rather,
the
CSFII
reports
an
adjusted
sample
weight
for
each
individual
who
reported
two
non
B
consecutive
days
of
consumption
data
during
the
survey.
Given
that
only
the
adjusted
weight
was
available,
and
not
the
specific
sample
and
population
size
in
each
phase,
it
was
necessary
to
estimate
the
mean
using
ratio
estimation
techniques
and
the
variance
of
the
mean
using
the
ultimate
cluster
methodology,
which
does
not
partition
the
variance
into
sample
design
components
(
refer
to
Appendix
D).

Interval
estimates
for
percentiles
are
bootstrap
intervals.
The
reported
bootstrap
intervals
do
not
result
from
direct
estimates
of
the
standard
deviation
of
the
point
estimate.
Rather,
the
bootstrap
estimates
result
from
the
percentile
method,
which
estimates
the
lower
and
upper
bounds
for
the
interval
estimate
by
the
100

percentile
and
100(
1
B

)
percentile
estimates
from
the
nonparametric
distribution
of
the
given
point
estimate.
This
distribution
of
the
observed
values
of
the
given
point
estimate
is
determined
from
repeated
resampling
of
the
empirical
data.
3­
7
3.3
Rounding
Procedures
Tabulated
estimates
of
per
capita
ingestion
in
milliliters
are
rounded
to
the
nearest
whole
number.

Conventional
rounding
procedures
were
applied
such
that
the
whole
number
remained
the
same
if
decimal
estimates
were
less
than
0.5
and
increased
by
one
if
the
decimal
estimate
was
0.5
or
greater.
Whole
number
presentations
do
not
reflect
significant
digits
as
the
number
of
significant
digits
is
not
available
for
the
CSFII.

3.4
Units
of
Measure
Including
Conversion
Factors
Per
capita
water
ingestion
estimates
are
presented
in
this
report
in
units
of
ml/
person/
day
or
ml/
kg
body
wt/
day.
The
person/
day
component
reflects
that
estimates
are
based
on
an
average
of
two
days
of
consumption.
When
the
units
are
ml/
kg
body
wt/
day,
the
average
water
ingestion
over
two
days
by
an
individual
is
divided
by
the
individual's
body
weight.
Body
weight
is
recorded
in
the
CSFII
in
pounds
(
lb).

These
pounds
are
converted
to
kilograms
by
multiplying
the
reported
body
weight
by
a
factor
of
0.454
kg/
lb.

Survey
participants
reported
the
amount
of
plain
water
ingested
directly
as
a
beverage
in
fluid
ounces.

Reported
ingestions
were
multiplied
by
29.574
to
convert
fluid
ounces
to
milliliters.
Water
ingested
indirectly
from
foods
with
water
added
at
home
or
locally
during
the
final
stage
of
preparation
was
estimated
in
grams
as
food
consumption
and
recipe
amount
are
reported
in
the
CSFII
in
grams.
These
grams
of
water
were
converted
into
milliliters
based
on
the
assumption
that
the
specific
gravity
of
water
is
one
for
the
temperature
range
of
ingested
foods.

3.5
Sample
Size
Criteria
Estimates
based
on
small
sample
sizes
may
be
less
statistically
reliable
than
estimates
based
on
larger
sample
sizes.
"
Third
Report
on
Nutrition
Monitoring
in
the
United
States"
suggest
minimal
reporting
requirements
(
LSRO
1995).
If
the
sample
size
is
less
than
30*(
variance
inflation
factor),
the
estimate
of
the
mean
may
be
unreliable
and
is
marked
with
an
asterisk.
If
the
(
sample
size)*(
1
 
percentile)
is
less
than
8*(
variance
inflation
factor),
then
the
percentile
estimate
may
be
unreliable
and
is
marked
with
an
asterisk.

The
variance
inflation
factor
for
the
two
days
of
CSFII
data
is
1.60.
The
variance
inflation
factor
is
sample
design
specific
and
is
a
broadly
calculated
design
effect
measure.
In
accordance
with
the
suggested
minimum
reporting
requirements,
mean
ingestions
estimated
with
sample
size
<
48
are
marked
with
an
asterisk
to
designate
that
they
may
be
statistically
unreliable.
Similarly,
percentiles
estimated
with
sample
size
<
12.8/(
1
 
percentile)
are
marked
and
may
be
statistically
unreliable.
3­
8
THIS
PAGE
INTENTIONALLY
LEFT
BLANK
4­
1
4.
RESULTS
This
chapter
presents
point
and
interval
estimates
of
the
mean,
90th
percentile
and
95th
percentile
for
select
subpopulations.
We
augment
tabulated
estimates
in
this
chapter
with
graphical
presentations
of
the
empirical
distributions
of
per
capita
water
ingestion
estimates
for
select
subpopulations.
Because
EPA
anticipates
that
per
capita
ingestion
of
community
water
will
be
of
primary
interest
to
the
readers
of
this
report,
we
emphasize
these
results
in
this
chapter.
Since
children
less
than
one
year
of
age
and
pregnant,

lactating,
and
childbearing
 
age
women
are
considered
to
be
high
risk
subpopulations,
we
also
discuss
their
estimates
in
this
chapter.
Finally,
to
reflect
changes
in
consumer
behavior
since
the
1977
 
78
survey,
which
was
the
basis
for
the
Ershow
and
Cantor
report,
we
report
per
capita
ingestion
of
bottled
water
from
1994
 
96
CSFII.

This
report
provides
tables
and
figures
of
per
capita
estimates
of
daily
average
water
ingestion.
Tables
and
figures
of
estimates
are
provided
for
all
individuals
and
for
"
consumers
only"
by
source,
and
type
of
ingestion.
Sources
of
ingestion
include
community
water,
bottled
water,
other
sources,
and
total
water
(
all
sources
combined).
Other
sources
include
a
household
well,
household
rain
cistern,
household
or
public
spring
and
other
sources.
Types
of
ingestion
are
direct
for
plain
water
ingested
as
a
beverage,
indirect
for
ingestion
of
the
water
added
to
foods
and
beverages
during
final
preparation
at
home
or
by
food
service
establishments
(
e.
g.,
school
cafeterias
and
restaurants),
and
both
direct
and
indirect
for
combined
direct
and
indirect
water
ingestion.
Biological
and
commercial
water
are
excluded
from
these
estimates
of
water
ingestion.
Estimates
are
provided
in
both
units
of
milliliters/
person/
day
(
ml/
person/
day)
and
milliliters/
kilogram
of
body
weight/
day
(
ml/
kg/
day).

Refer
to
Appendix
E
for
a
more
comprehensive
set
of
empirical
distributions
of
estimated
per
capita
water
ingestion.
In
addition
to
the
broad
age
categories
reported
in
this
chapter,
Appendix
E
provides
estimates
of
water
ingestion
by
finer
age
categories.
Appendix
E
also
provides
a
more
extensive
percentile
distribution
which
includes
point
estimates
of
the
mean,
and
1st,
5th,
10th,
25th,
50th,
75th,
90th,
95th
and
99th
percentiles.

It
should
be
noted
that
the
dispersion
of
individuals
for
some
subpopulations
across
CSFII
estimation
strata
did
not
always
support
generation
of
variance
estimates.
Therefore,
the
following
discussions
will
point
out
differences
in
mean
per
capita
ingestion
between
subpopulations,
but
these
differences
are
a
quantitative
statement
and
do
not
imply
statistical
differences.
Without
variance
estimates
about
the
means
of
the
subpopulations,
we
cannot
perform
formal
statistical
tests
to
ascertain
whether
means
for
various
subpopulations
differ
statistically.
4­
2
4.1.
a
Ingestion
of
Community
Water
(
ml/
person/
day)

The
mean
daily
average
of
estimated
per
capita
community
water
is
927
ml/
person.
This
average
projects
to
the
population
of
the
United
States.
The
90%
confidence
interval
about
the
mean
is
902
to
951
ml/
person/
day.
The
estimated
90th
percentile
from
the
empirical
distribution
of
daily
average
per
capita
community
water
ingestion
is
2.016
liters.
A
90%
bootstrap
interval
about
the
90th
percentile
intake
estimate
for
all
individuals
is
1.991
liters
to
2.047
liters
(
See
Table
4
B
1
B
A).

Figure
4
B
1
B
F1
depicts
the
empirical
distribution
of
daily
average
per
capita
community
water
ingestion
for
all
individuals
during
1994
B
96.
Considering
that
the
5th
percentile
estimate
from
this
empirical
distribution
is
zero
and
the
95th
percentile
estimate
is
2.544
liters
(
See
Appendix
E,
Part
I,
Table
A1),
the
empirical
distribution
is
obviously
skewed.
That
is,
the
mean
estimate
is
influenced
by
people
ingesting
either
zero
to
a
very
little
amount
of
water
or
very
large
volumes
of
water.
Figure
4
B
1
B
G1
displays
a
histogram
of
daily
average
per
capita
community
water
estimates.
This
histogram
illustrates
that
most
of
the
daily
average
ingestion
reported
by
CSFII
respondents
are
less
than
two
liters.
The
bar
with
the
midpoint
of
1.575
liters
has
an
upper
value
of
2.2
liters.
This
is
between
the
estimated
90th
and
95th
percentiles
of
the
empirical
distribution.

The
mean
daily
average
of
estimated
per
capita
ingestion
of
community
water
for
"
consumers
only"
is
1.0
liter/
person
(
90%
confidence
interval
is
976
ml
to
1.024
liter/
person).
The
estimated
90th
percentile
from
the
empirical
distribution
of
daily
average
per
capita
community
water
ingestion
for
consumers
is
2.069
liters
(
90%
bootstrap
interval
is
2.041
to
2.106
liters/
person/
day)
(
See
Table
4
B
2
B
A).
Figure
4
B
2
B
F1
portrays
the
empirical
distribution
of
daily
average
per
capita
community
water
ingestion
for
consumers.

One
point
of
clarification
regarding
the
histograms
(
See
Figures
4
B
1
B
G1,
4
B
1
B
G2,
4
B
2
B
G1,
and
4
B
2
B
G2)
is
necessary.
Amounts
printed
along
the
x
B
axis
are
midpoint
values
for
all
the
bars
except
the
first
two.
That
is,
the
width
of
the
third
bar
and
beyond
is
630
ml.
The
first
bar
for
Figures
4
B
1
B
G1
and
4
B
1
B
G2
represents
nonconsumers
(
respondents
with
zero
reported
water
ingestion
or
with
missing
ingestion)
and
respondents
with
minimal
ingestion.
In
this
case,
minimal
ingestion
is
more
than
zero
but
less
than
157.5
ml/
day.
The
second
bar
represents
ingestion
of
greater
than
157.5
ml/
day
but
less
than
630
ml/
day.
Therefore,
the
first
two
bars
together
represent
ingestion
of
630
ml/
day
or
less.
All
other
bars
are
each
intervals
of
630
ml/
day.
Bars
are
defined
similarly
for
Figures
4
B
2
B
G1
and
4
B
2
B
G2
for
consumers
except
that
the
first
bar
only
represents
respondents
with
minimal
ingestion
defined
as
more
than
zero
but
less
than
157.5
ml/
day.

4.1.
b
Ingestion
of
Community
Water
(
ml/
kg
of
body
weight/
day)

The
mean
ingestion
of
community
water
for
the
United
States
population,
reported
in
units
of
per
kilogram
of
body
weight,
is
16
ml/
kg/
day
(
90%
confidence
interval
is
15
to
16
ml/
kg/
day)
(
See
Table
4­
3
4
B
1
B
B2).
For
"
consumers
only,"
the
mean
ingestion
of
community
water
is
17
ml/
kg/
day
(
90%
confidence
interval
is
16
to
17
ml/
kg/
day)
(
See
Table
4
B
2
B
B2).
The
90th
percentile
from
the
empirical
distribution
of
daily
average
per
capita
ingestion
of
community
water
for
all
individuals
and
"
consumers
only"
is
33
ml/
kg
(
See
Table
4
B
1
B
B2
and
Table
4
B
2
B
B2).

4.2.
a
Ingestion
of
Community
Water
by
Age
and
Gender
(
ml/
person/
day)

In
the
United
States
population,
individuals
20
years
and
older
ingest
an
average
of
1.098
liters
(
90%

confidence
interval
is
1.068
to
1.127
liters)
of
community
water
per
day.
This
is
followed
by
individuals
11
to
19
years
old
who
ingest
an
average
of
683
ml
daily
(
90%
confidence
interval
is
634
to
732
ml),

children
one
to
ten
years
old
who
ingest
an
average
of
400
ml
daily
(
90%
confidence
interval
is
380
to
420
ml),
and
children
less
than
one
year
old
who
ingest
an
average
of
342
ml
daily
(
90%
confidence
interval
is
295
to
388
ml)
(
See
Table
4
B
1
B
B1).

Results
for
"
consumers
only"
by
age
category
are
similar.
Individuals
20
years
or
older
ingest
an
average
of
1.176
liters
of
community
water
per
day
(
90%
confidence
interval
is
1.148
to
1.204
liters/
day).

Young
adults
11
to
19
years
old
ingest
an
average
of
735
ml/
day
(
90%
confidence
interval
is
684
to
786
ml/
day),
children
one
to
ten
years
old
ingest
an
average
of
435
ml/
day
(
90%
confidence
interval
is
414
to
457
ml/
day),
and
children
less
than
one
year
old
ingest
an
average
of
513
ml/
day
(
90%
confidence
interval
is
460
to
567
ml/
day)
(
See
Table
4
B
2
B
B1).

The
mean
community
water
ingested
by
males
is
higher
than
that
ingested
by
females
in
all
age
categories
except
for
children
younger
than
one
year
old
and
children
one
to
ten
years
of
age.
The
highest
mean
per
capita
ingestion
by
males
is
found
in
the
20
years
and
older
age
group.
The
mean
for
this
group
is
1.162
liters/
person/
day
and
the
90th
percentile
is
2.337
liters/
person/
day.
For
females
20
years
and
older,
the
mean
daily
average
of
estimated
per
capita
community
water
ingestion
is
1.039
liters,
while
the
90th
percentile
estimate
is
2.126
liters.
Ninetieth
percentile
estimates
are
less
than
two
liters/
person/
day
for
males
and
females
less
than
one
year
old,
between
the
ages
of
one
and
ten,
and
between
the
ages
of
11
and
19
(
See
Table
4
 
1
 
C1).

Similarly,
male
consumers
ingest
more
community
water
on
average
than
female
consumers.
Male
consumers
20
years
and
older
have
the
highest
mean
per
capita
ingestion
(
1.242
liters).
The
90th
percentile
estimate
of
daily
average
per
capita
community
water
ingestion
for
male
consumers
20
years
and
older
is
2.387
liters.
The
daily
average
per
capita
community
water
ingestion
for
female
consumers
20
years
and
older
is
1.116
liters,
and
the
90th
percentile
estimate
is
2.165
liters.
The
mean
difference
between
the
two
genders
for
individuals
20
years
and
older
is
126
ml
or
4.2
fluid
ounces.
Ninetieth
percentile
estimates
are
less
than
two
liters/
person/
day
for
male
and
female
consumers
for
all
age
categories
younger
than
20
years
old
(
See
Table
4
 
2
 
C1).
4­
4
4.2.
b
Ingestion
of
Community
Water
By
Age
and
Gender
(
ml/
kg
of
body
weight/
day)

For
all
individuals,
the
lowest
mean
daily
average
per
capita
ingestion
from
community
water,
reported
per
kilogram
of
body
weight,
is
12
ml/
kg
for
individuals
aged
11
to
19
years
old.
The
highest
mean
daily
average
per
capita
ingestion
is
46
ml/
kg
for
children
less
than
one
year
old.
Adults
20
years
and
older
have
a
mean
daily
average
per
capita
ingestion
of
15
ml/
kg,
and
children
one
to
ten
years
old
have
a
mean
daily
average
per
capita
ingestion
of
19
ml/
kg
(
See
Table
4
 
1
 
B2).
This
pattern
is
similar
for
consumers
(
See
Table
4
 
2
 
B2).
Thus,
based
on
per
kilogram
body
weight,
the
infants
less
than
one
year
of
age
consume
approximately
three
times
the
estimated
amount
of
community
water
as
the
adult
20
years
or
older.

Males
and
females
in
the
U.
S.
population
have
similar
mean
daily
average
per
capita
ingestion,

reported
per
kilogram
of
body
weight,
from
community
water.
Females
have
higher
mean
ingestion
for
all
age
groups
except
for
individuals
11
to
19
years
old
(
11
ml/
kg/
day
vs.
13
ml/
kg/
day)
(
See
Table
4
 
1
 
C2).

The
comparison
between
mean
ingestion
estimates
for
male
and
female
consumers
is
similar
(
See
Table
4
 
2
 
C2).

4.3.
a
Ingestion
of
Community
Water
for
Children
Younger
Than
One
Year
of
Age
(
ml/
person/
day)

The
age
group
with
the
lowest
mean
ingestion
of
direct
and
indirect
community
water
for
both
genders
is
children
less
than
one
year
old.
This
is
also
the
only
age
group
where
the
mean
per
capita
ingestion
by
females
(
384
ml)
is
higher
than
that
for
males
(
298
ml)
(
See
Table
4
 
1
 
C1).
Similarly,
female
consumers
less
than
one
year
old
have
a
higher
mean
per
capita
ingestion
of
community
water
than
male
consumers
(
560
ml/
day
vs.
462
ml/
day)
(
See
Table
4
 
2
 
C1).

4.3.
b
Ingestion
of
Community
Water
for
Children
Younger
Than
One
Year
of
Age
(
ml/
kg
of
body
weight/
day)

Children
younger
than
one
year
old
have
a
mean
intake
of
community
water
of
46
ml/
kg/
day,
the
highest
of
the
age
categories.
The
average
for
all
individuals
(
all
ages)
is
16
ml/
kg/
day
(
See
Table
4
 
1
 
B2)
Likewise,
consumers
less
than
one
year
old
have
the
highest
mean
ingestion,
69
ml/
kg/
day.
The
mean
for
all
individuals
(
all
ages)
is
17
ml/
kg/
day.
Therefore,
infants
younger
than
one
year
of
age
ingest
approximately
three
to
four
times
the
estimated
amount
of
community
water
than
do
individuals
in
all
age
groups
(
See
Table
4
B
2
B
B2).

4.4.
a
Ingestion
of
Community
Water
for
Women
in
Childbearing
Years
(
ml/
person/
day)

Lactating
women
have
the
highest
mean
water
ingestion.
The
mean
daily
average
ingestion
by
lactating
women
is
1.379
liters,
while
the
means
for
pregnant
women
and
women
in
childbearing
years
are
819
and
922
ml,
respectively.
The
75th
percentile
estimate
for
lactating
women
exceeds
two
liters
(
2.263
4­
5
liters),
compared
to
1.272
liters
for
women
of
childbearing
age.
The
90th
percentile
daily
average
per
capita
ingestion
by
lactatingwomen
is
2.872
liters
as
compared
to
2.008
liters
for
women
in
childbearing
years.
The
90th
percentile
estimate
from
the
empirical
distribution
of
daily
average
per
capita
ingestion
of
community
water
by
pregnant
women
is
1.816
liters.
The
95th
percentile
estimates
from
the
empirical
distributions
of
daily
average
per
capita
ingestion
of
community
water
for
pregnant
women,
lactating
women,
and
women
aged
15
to
44
are
2.501,
3.434,
and
2.604
liters,
respectively
(
See
Table
4
B
1
B
E
and
Appendix
E,
Part
I,
TableA3).

Similarly,
for
"
consumers
only,"
lactating
women
ingest
more
water
than
do
pregnant
women
or
women
in
the
childbearing
ages.
The
mean
daily
average
ingestion,
for
"
consumers
only,"
of
community
water
for
lactating
women
is
1.665
liters,
for
pregnant
women
is
872
ml,
and
for
women
in
childbearing
years
is
984
ml.
As
noted
above,
the
75th
percentile
estimate
of
ingestion,
for
"
consumers
only,"
for
lactating
women
exceeds
two
liters/
day
(
2.417
liters/
day),
compared
to
1.314
liters/
day
for
women
of
childbearing
age
and
1.424
liters/
day
for
pregnant
women.
The
90th
percentile
estimate
of
ingestion
for
"
consumers
only"
exceeds
two
liters/
day
for
both
lactating
women
and
women
of
childbearing
age
(
2.959
liters
and
2.044
liters,
respectively).
The
95th
percentile
estimates
of
daily
average
per
capita
ingestion
of
community
water
for
pregnant
consumers,
lactating
consumers,
and
female
consumers
aged
15
to
44
are
2.588,
3.588,
and
2.722
liters,
respectively
(
See
Table
4
B
2
B
E
and
Appendix
E,
Part
III,
Table
A3).

4.4.
b
Ingestion
of
Community
Water
for
Women
in
Childbearing
Years
(
ml/
kg
of
body
weight/
day)

When
estimates
are
reported
for
all
women
of
childbearing
age
in
units
of
milliliter/
kilogram
of
body
weight/
day,
the
mean
ingestion
by
lactating
women
is
the
highest
at
21
ml/
kg.
Pregnant
women
have
the
lowest
mean
ingestion
of
community
water
with
a
mean
of
13
ml/
kg.
Women
in
childbearing
years
have
an
estimated
mean
ingestion
of
14
ml/
kg.
The
90th
and
95th
percentiles
from
the
empirical
distribution
of
daily
average
per
capita
ingestion
per
kilogram
of
body
weight
for
lactating
women
both
exceed
50
ml/
kg.

Ninetieth
percentile
estimates
for
pregnant
women
and
women
in
childbearing
years
are
32
ml/
kg,
while
the
95th
percentile
estimates
from
these
two
distributions
are
43
and
39
ml/
kg,
respectively
(
See
Appendix
E,

Part
II,
Table
A3).

The
mean
ingestion
of
community
water
is
26
ml/
kg/
day
for
lactating
consumers,
14
ml/
kg/
day
for
pregnant
consumers,
and
15
ml/
kg/
day
for
female
consumers
aged
15
to
44.
The
90th
and
95th
percentiles
of
daily
average
per
capita
ingestion
per
kilogram
of
body
weight
for
lactating
consumers
both
exceed
50
ml/
kg.
Ninetieth
percentile
estimates
for
pregnant
consumers
and
female
consumers
aged
15
to
44
are
33
and
32
ml/
kg/
day,
while
the
95th
percentile
estimates
from
these
two
groups
are
43
and
39
ml/
kg/
day
(
See
Appendix
E,
Part
IV,
Table
A3).
4­
6
4.5
Ingestion
of
Bottled
Water
and
Water
from
Other
Sources
Mean
per
capita
ingestion
of
bottled
water
for
the
United
States
population
is
161
ml.
The
90th
and
95th
percentile
estimates
from
the
empirical
distribution
of
daily
average
per
capita
ingestion
of
bottled
water
for
the
United
States
population
are
591
ml
and
1.036
liters,
respectively
(
See
Table
4
B
1
B
A).
Mean
ingestion
of
water
from
other
sources
by
the
United
States
population
is
128
ml.
Ninetieth
and
95th
percentile
estimates
of
per
capita
ingestion
of
water
from
other
sources
are
343
ml
and
1.007
liters,

respectively
(
See
Table
4
B
1
B
A).
Other
sources
include
water
from
wells,
rain
cisterns,
springs,
and
sources
identified
by
respondents
as
"
other."
Comparing
the
mean
daily
average
per
capita
ingestion
of
bottled
water
and
water
from
other
sources
to
total
water
ingestion
regardless
of
sources
(
1.232
liters)

suggests
that
13
percent
of
total
water
ingestion
is
attributable
to
bottled
water
while
10
percent
is
attributable
to
water
from
other
sources.
Community
water
comprises
75
percent
of
the
total
water
ingestion
by
individuals
in
the
United
States
population.

Daily
average
per
capita
ingestion
for
consumers
of
bottled
water
is
737
ml/
person.
The
90th
and
95th
percentile
estimates
of
ingestion
for
consumers
of
bottled
water
are
1.568
liters/
person/
day
and
1.967
liters/
person/
day
(
See
Table
4
B
2
B
A).
The
daily
average
per
capita
ingestion
for
consumers
of
water
from
other
sources
is
965
ml.
The
90th
and
95th
percentile
estimates
of
daily
per
capita
ingestion
are
1.971
and
2.475
liters
(
See
Table
4
B
2
B
A).

4.6.
a
Ingestion
of
TotalWater
for
All
Individuals
The
mean
estimate
of
total
water
ingestion
(
ingestion
of
water
from
all
sources)
for
the
general
population
is
1,232
ml/
person/
day
(
90%
confidence
interval
is
1,199
to
1,265
ml/
person/
day).
The
90th
and
95th
percentiles
of
the
distribution
are
2,341
ml/
person/
day
and
2,908
ml/
person/
day,
respectively
(
See
Table
4
 
1
 
D1).
Approximately
84
percent
of
the
U.
S.
population
ingests
two
liters
or
less
per
day
of
total
water
(
See
Figure
4
 
1
 
F2).

For
babies
younger
than
one
year
old,
the
estimated
mean
consumption
of
total
water
is
484
ml/
person/
day
(
90%
confidence
interval
is
438
to
530
ml/
person/
day).
The
90th
and
95th
percentiles
of
consumption
are
949
ml/
person/
day
(
90%
bootstrap
interval
is
893
to
1,046
ml/
person/
day)
and
1,182
ml/
person/
day,
(
90%
bootstrap
interval
is
1,046
to
1,282
ml/
person/
day),
respectively.
The
mean
value
of
the
daily
total
water
ingestion
for
a
child
one
to
ten
years
old
is
528
ml/
person/
day
(
90%
confidence
interval
is
505
to
552
ml/
person/
day).
The
90th
and
95th
percentiles
of
total
water
ingestion
are
1,001
ml/
person/
day
(
90%
bootstrap
interval
is
980
to
1,027
ml/
person/
day)
and
1,242
ml/
person/
day
(
90%

bootstrap
interval
is
1,189
to
1,264
ml/
person/
day),
respectively
(
See
Table
4
 
1
 
D1).
Thus,

approximately
90
percent
of
the
children
ten
years
of
age
or
younger
consume
less
than
or
equal
to
the
standard
default
value
of
one
liter
per
day.
4­
7
4.6.
b
Ingestion
of
TotalWater
for
"
Consumers
Only"

The
most
conservative
water
ingestion
distributions
for
the
two
 
day
average
per
capita
ingestion
of
water
from
all
sources
are
by
"
consumers
only".
The
estimated
mean
for
the
general
population
is
1,241
ml/
person/
day
(
90%
confidence
interval
is
1,208
to
1,274
ml/
person/
day).
The
90th
and
95th
percentiles
are
2,345
ml/
person/
day
(
90%
bootstrap
interval
is
2,315
to
2,378
ml/
person/
day)
and
2,922
ml/
person/
day
(
90%
bootstrap
interval
is
2,848
to
2,959
ml/
person/
day),
respectively
(
See
Table
4
 
2
 
D1).

Approximately
83
percent
of
"
consumers
only"
ingest
less
than
or
equal
to
the
standard
two
liters/
day
when
considering
total
water
(
See
Fig.
4
 
2
 
F2).
For
"
consumer
only"
infants
younger
than
one
year
old,
the
estimated
mean
ingestion
of
total
water
is
563
ml/
person/
day
(
90%
confidence
interval
is
508
to
618
ml/
person/
day).
The
90th
and
95th
percentiles
are
968
ml/
person/
day
(
90%
bootstrap
interval
is
940
to
1,121
ml/
person/
day)
and
1,236
ml/
person/
day
(
90%
bootstrap
interval
is
1,121
to
1,282
ml/
person/
day),

respectively.
For
"
consumer
only"
children
one
to
ten
years
of
age,
the
estimated
mean
consumption
of
total
water
is
532
ml/
person/
day
(
90%
confidence
interval
is
509
to
556
ml/
person/
day).
The
90th
and
95th
percentiles
are
1,004
ml/
person/
day
(
90%
bootstrap
interval
is
980
to
1,030
ml/
person/
day)
and
1,242
ml/
person/
day
(
90%
bootstrap
interval
is
1,198
to
1,284
ml/
person/
day),
respectively
(
See
Table
4
 
2
 
D2).

Therefore,
even
by
the
most
conservative
estimate
(
i.
e.,
water
from
all
sources
and
excluding
the
zero
consumers),
90%
of
all
children
ten
years
or
younger
drink
less
than
or
equal
to
the
default
value
of
one
liter
of
water
per
day.
4­
8
THIS
PAGE
INTENTIONALLY
LEFT
BLANK
5­
1
5.
DISCUSSION
All
surveys
have
strengths
and
limitations
when
assessed
against
the
specific
objective
being
addressed.

There
are
also
biases
introduced
in
the
survey
process.
Section
5.1
presents
the
strengths
and
limitations
of
the
USDA's
1994
B
96
CSFII
data
for
supporting
the
estimates
reported
in
this
document.
Section
5.2
identifies
and
discusses
sources
of
bias
and
error
in
the
1994
B
96
CSFII
with
respect
to
water
ingestion
estimates.
Section
5.3
presents
the
report
conclusions,
and
Section
5.4
provides
a
listing
of
references
used
in
this
report.

5.1
Survey
Strengths
and
Limitations
The
strengths
of
the
USDA's
1994
B
96
CSFII
survey
for
supporting
estimates
of
per
capita
water
ingestion
are
twofold.
First,
the
survey
design
is
structured
to
obtain
a
statistically
representative
sample
of
the
United
States
population.
Second,
the
survey
is
designed
to
record
daily
intakes
of
foods
and
nutrients
and
support
estimation
of
food
consumption.
These
features
are
in
direct
alignment
with
the
objective
of
producing
current,
per
capita
water
ingestion
estimates
for
the
United
States
population
and
for
population
subsets
sensitive
to
potential
contaminants
in
drinking
water.

The
1994
B
96
CSFII
survey
design
allows
the
combination
of
three
years
of
data
through
a
weighting
scheme.
This
combination
of
three
years
provides
a
sample
of
over
15,000
respondents.
With
increased
sample
sizes,
the
precision
and
accuracy
of
estimates
are
improved
and
the
support
for
subpopulation
estimates
is
enhanced.
This
design
structure,
in
conjunction
with
the
implementation
of
a
sampling
protocol,
increases
the
sample's
representation
of
the
United
States
population
and
minimizes
seasonal
and/
or
regional
bias
from
respondents.
Low
B
income
individuals
are
oversampled
to
ensure
their
representation
in
the
survey.
Finally,
the
survey
weight
associated
with
each
respondent's
information
to
project
the
response
to
the
population
has
been
adjusted
for
nonresponse
bias.
These
adjustments
were
based
on
sociodemographic
factors.
Nonresponse
adjustments
were
also
significantly
reduced
for
the
current
CSFII.
The
response
rate
for
participants
with
multiple
days
of
food
intake
information
is
75.9
percent
for
the
1994
B
96
CSFII,
as
opposed
to
approximately
45
percent
for
the
1989
B
91
CSFII.

The
method
employed
to
collect
dietary
intake
data
also
strengthened
the
CSFII
design
for
supporting
per
capita
ingestion
estimates.
For
example,
the
USDA's
1994
B
96
CSFII
survey
was
administered
by
an
interviewer
on
both
days
of
data
collection.
This
administration
provided
multiple
passes
through
the
day's
intake
to
facilitate
more
complete
responses.
Previous
surveys
have
relied
on
interviewer
administration
for
the
first
day
and
self
B
administration
on
subsequent
days.
This
change
in
administration
method
insures
consistency
with
respect
to
the
way
responses
are
recorded
across
interview
days.
5­
2
Previous
CSFII
surveys
have
collected
dietary
intake
information
on
consecutive
days.
This
collection
method
raises
issues
about
the
contribution
of
within
B
individual
variance
to
overall
estimates.
Because
the
1994
B
96
CSFII
collects
data
on
two
non
B
consecutive
days,
the
within
B
individual
variance
component
is
diminished.
The
third
change
in
data
collection
methods
that
facilitates
completion
of
the
objective
of
this
report
is
that
previous
surveys
included
all
members
of
a
household
in
the
survey.
The
1994
B
96
survey
includes
a
subsample
of
household
members
with
sampling
rates
varying
to
achieve
more
responses
from
children
and
the
elderly.

Another
important
feature
of
the
1994
B
96
CSFII
that
supports
per
capita
estimation
of
water
ingestion
is
the
questionnaire
design.
The
questionnaire
collects
data
on
a
household's
source
of
drinking
water
and
water
used
for
the
preparation
of
foods
and
beverages.
It
also
allows
a
respondent
to
indicate
if
water
was
ingested
at
home
or
away
from
home.
This
information
directly
supports
the
assignment
of
water
source
for
both
direct
and
indirect
water
intake.
The
1994
B
96
CSFII
Technical
Support
Files
supported
the
estimation
of
the
amount
of
water
ingested
through
food.
This
enhances
the
estimation
of
indirect
water
ingestion
and
partitions
it
from
water
directly
ingested
as
a
beverage.

The
limitations
of
the
CSFII
survey
for
supporting
per
capita
ingestion
estimates
involve
the
length
of
time
data
were
collected,
the
influence
of
extreme
values
on
estimates,
and
the
availability
of
information
to
support
variance
estimation.
The
CSFII
survey
collects
only
two
non
B
consecutive
days
of
data.
Because
daily
averages
are
estimated
from
each
respondent
from
only
two
days,
the
precision
of
an
individual's
daily
average
consumption
is
diminished.
Also,
the
limited
time
period
of
dietary
intake
collection
does
not
produce
usual
intake
estimates.
Usual
intakes
are
defined
as
"
the
long
run
average
of
daily
intakes
of
a
dietary
component
by
an
individual."
Rather,
the
estimates
presented
in
this
report
characterize
the
empirical
distribution
of
daily
average
per
capita
ingestion.
Because
the
data
from
the
CSFII
are
not
usual
intakes
and
some
consumers
report
no
direct
and
minimal
amounts
of
indirect
water
ingestion,
while
other
consumers
report
over
two
liters
of
ingestion,
the
empirical
distribution
of
daily
average
per
capita
ingestion
can
be
skewed.

Another
limitation
of
the
1994
B
96
CSFII
is
a
function
of
the
way
that
survey
data
are
reported.
Data
from
two
variance
estimation
units
are
required
to
generate
an
estimate
of
the
variance
within
a
variance
estimation
stratum.
These
variances
are
then
summed
across
strata
to
generate
a
variance
estimate
for
the
subpopulation.
For
many
of
the
subpopulations
evaluated
in
this
report,
numerous
strata
did
not
have
information
for
two
variance
estimation
units.
Because
there
is
insufficient
information
in
the
naming
convention,
combining
data
across
like
strata
was
not
possible.
Therefore,
the
survey
did
not
support
variance
estimation
for
many
of
the
reported
subpopulations.
Because
of
this,
means
differences
cannot
be
formally
tested
and
interval
estimates
about
the
mean
and
upper
percentiles
cannot
be
supplied,
except
for
the
larger
subpopulation.
All
reported
differences
are
empirical
as
opposed
to
statistical.
Also,
certain
variables,
such
as
region,
are
at
a
summary
level.
USDA
has
named
the
States
within
a
region.
Estimates
by
State,
however,
are
not
trackable
because
USDA
data
do
not
contain
a
variable
identifying
States.
For
this
reason,
water
ingestion
estimates
by
State
are
not
possible.
5­
3
Statistically
significant
differences
can
be
found
by
comparing
the
confidence
intervals
between
two
independent
groups.
If
the
confidence
intervals
for
the
two
groups
do
not
overlap,
then
the
estimates
for
these
groups
are
statistically
significant
at
the
0.10
alpha
level
since
90%
confidence
intervals
are
reported.

For
example,
children
one
to
ten
years
old
(
90%
confidence
interval
about
the
mean
is
380
to
420
ml/
person/
day)
ingest
significantly
less
community
water
than
children
11
to
19
years
old
(
90%
confidence
interval
about
the
mean
is
634
to
732
ml/
person/
day)
(
See
Table
4
 
1
 
B1).

A
final
limitation
is
that
the
survey
does
not
support
water
ingestion
estimates
for
subpopulations
with
different
lifestyles,
occupations,
or
activities.
Examples
include:


People
with
traditional
life
styles
(
e.
g.,
Native
Americans
and
recent
immigrants).


People
who
live
in
hot
climate
areas.


People
who
consume
large
amounts
of
water
because
of
physical
activity.


People
with
health
conditions
that
affect
water
ingestion,
such
as
diabetes,
kidney
disease,
conditions
requiring
rapid
rehydration
needs
(
GI
upsets,
food
poisoning),
and
disorders
of
water
and
sodium
metabolism.

While
individuals
from
these
specific
subpopulations
are
included
in
the
survey
and
U.
S.
population
estimates,
they
were
not
targeted
during
survey
design
and
thus
do
not
occur
in
high
enough
frequencies
to
support
estimate
generation.

5.2
Sources
of
Error,
Bias,
and
Uncertainty
All
surveys
contain
errors
despite
the
diligence
of
the
design
statistician
and
the
respondents.
These
errors
ultimately
lead
to
bias
and
uncertainty
in
the
estimates
resulting
from
the
survey's
data.
Some
errors
are
quantifiable,
while
others
are
not.
Random
error
occurs
in
all
stochastic
processes.
To
quantify
error
and
bias,
we
must
know
the
true
population
value.
In
reality,
these
are
not
known.
In
general,
the
estimation
process
assumes
that
the
true
population
value
is
known
and
the
error
is
random
or
partitioned
to
assess
components
of
the
variance.
In
complex
surveys,
these
assumptions
may
be
violated.

In
general,
there
are
three
sources
of
error
in
a
survey.
Two
of
these
sources
involve
the
survey
design
and
data
collection.
The
third
source
of
error
is
introduced
during
the
use
of
the
data.
The
following
paragraphs
discuss
these
sources
of
error
specific
to
the
1994
B
96
USDA's
CSFII
survey
and
its
use
to
generate
the
estimates
presented
in
this
report.

The
first
source
of
error
is
attributed
to
the
survey
design.
All
designs
are
constructed
to
minimize
the
coefficient
of
variation
with
respect
to
a
given
parameter.
For
the
1994
B
96
CSFII,
the
goal
was
to
5­
4
minimize
the
variance
of
the
mean
Day
1
saturated
fat
and
iron
intakes.
In
this
report,
we
address
water
ingestion.
Thus,
the
design
has
not
been
specifically
structured
to
minimize
the
coefficient
of
variation
with
respect
to
water
ingestion.
Another
design
error
is
attributed
to
nonresponse
and
the
representative
nature
of
the
sample
frame.
The
CSFII
adjusts
for
these
through
its
sample
weights.
The
method
USDA
used
to
derive
survey
sample
weights
is
discussed
in
Chapter
2
of
this
report.
For
the
combined
three
B
year
sample,

the
USDA
estimates
the
variance
inflation
factor
(
VIF)
for
two
B
day
respondents
to
be
1.60.
The
1994
B
96
CSFII
documentation
describes
the
VIF
as
"
the
proportional
increase
in
the
variance
of
survey
estimates
resulting
from
the
variation
in
weights"
and
indicates
that
the
VIF
measures
"
the
broadly
calculated
average
design
effect"
(
CSFII
survey
documentation,
p.
5
B
4
and
5
B
5).

The
second
source
of
error
is
measurement
error.
For
the
CSFII,
this
error
presents
itself
in
the
records
of
foods
and
beverages
ingested
by
the
participant.
Measurement
error
in
this
case
is
comprised
of
the
amount
of
a
food
or
beverage
consumption
reported
and
the
completeness
of
the
reported
consumption
record.
It
is
generally
anticipated
that
food
and
beverage
intakes
are
under
B
reported
(
Swan,
1983).

The
third
source
of
error
is
introduced
when
data
are
used.
The
first
incidence
of
this
occurring
is
in
the
data
coding
and
database
building
by
the
USDA.
Other
sources
occur
during
applications
of
data
conventions.
As
indicated
in
Chapter
3
of
this
report,
assumptions
were
made
about
sources
of
water
and
about
which
foods
were
prepared
at
home
or
by
a
food
service
establishment.

5.3
Conclusion
The
purpose
of
this
study
is
to
provide
current
estimates
of
per
capita
water
ingestion
in
the
United
States.
Results
are
presented
for
the
general
U.
S.
population
and
for
certain
sub
 
populations
(
i.
e.
gender
and
age
categories,
pregnant
and
lactating
women).
The
data
on
water
ingestion
were
obtained
from
the
U.
S.
Department
of
Agriculture's
1994­
96
Continuing
Survey
of
Food
Intake
by
Individuals.
The
estimates
report
mean
and
percentiles
from
empirical
distributions
for
both
direct
(
plain
water
ingested
as
a
beverage)
and
indirect
water
(
water
added
to
food
and
beverages
during
preparation).
Commercial
and
natural
water
in
the
food
and
beverages
are
not
included
in
the
analysis.

Two
liters/
person/
day
has
been
used
as
the
default
value
for
water
ingestion
by
EPA,
other
Federal
agencies,
and
theWHO.
This
value
is
supported
by
the
National
Cancer
Institute's
report
(
Ershow
and
Cantor,
1989)
based
on
1977
 
78
survey
data.
The
two
liters
included
the
sum
of
direct
and
indirect
tap
water
ingestion
and
was
the
88th
percentile
for
the
United
States
population
when
excluding
pregnant
and
lactating
women
and
breast
 
fed
children.

This
analysis,
based
on
1994
 
96
CSFII
data,
found
that
90
percent
of
the
population
of
the
United
States
ingests
two
liters/
day
or
less
of
community
water.
This
analysis
also
found
that
approximately
83
percent
of
the
population
ingests
two
liters/
day
or
less
of
total
water
(
i.
e.,
water
from
all
sources)
(
See
Figure
4
 
1
 
F2).
5­
5
For
babies
younger
than
one
year
of
age
who
ingested
community
water
during
the
two
survey
days
(
i.
e.,
"
consumers
only"),
this
analysis
showed
that
90
percent
ingested
less
than
or
equal
to
one
liter/
day
of
community
water.
For
babies
who
ingested
water
from
any
source
during
the
two
survey
days,
this
analysis
showed
that
over
90
(
but
less
than
95)
percent
ingested
less
than
or
equal
to
one
liter/
day
of
total
water.

When
considering
water
ingestion
in
units
of
milliliters
per
kilogram
of
body
weight
per
day,
this
analysis
shows
that
the
mean
per
capita
ingestion
rates
for
babies
younger
than
one
year
are
estimated
to
be
three
to
four
times
higher
than
the
mean
rates
for
the
population
as
a
whole.

Our
results
show
that
pregnant
women
do
not
differ
significantly
in
their
water
intake
compared
with
women
of
childbearing
age
(
age
15
 
44).
However,
lactating
women
ingest
significantly
more
water
than
the
other
two
groups.
These
conclusions
are
a
result
of
comparing
the
confidence
intervals
among
the
three
groups
of
women.
Note,
however,
that
the
pregnant
women
and
lactating
women
are
included
in
the
larger
group
of
childbearing
 
age
women.

The
mean
community
water
ingested
by
males
is
significantly
higher
than
that
ingested
by
females
in
all
age
categories
except
for
babies
younger
than
one
year
old
and
children
one
to
ten
years
of
age.
The
highest
mean
per
ingestion
by
males
is
found
in
males
in
the
20
years
and
older
age
group.

A
comparison
of
ingestion
by
various
sources,
indicates
that
community
water
comprises
75
percent
of
the
total
water
ingested
by
individuals
in
the
United
States
population,
followed
by
bottled
water
which
constitutes
13
percent
of
total
water
ingested
while
10
percent
is
attributable
to
water
from
other
sources.

The
results
presented
may
be
used
in
risk
assessment
analyses
where
exposures
that
occur
through
ingestion
of
water
are
of
concern.
The
ingestion
estimates
presented
provide
the
basis
for
evaluation
of
the
proportion
of
the
population
that
may
be
affected
under
various
exposure
scenarios.
5­
6
5.4
References
CSFII
survey
documentation,
file
1.
See
USDA/
ARS,
1998.

DHHS
(
U.
S.
Department
of
Health
and
Human
Services).
1996.
1996
Poverty
Guidelines.
Federal
Register
61(
43):
8286
 
8.

Efron,
Bradley.
1982.
The
Jackknife,
The
Bootstrap,
and
Other
Resampling
Plans.
Philadelphia,

Pennsylvania:
Society
for
Industrial
and
Applied
Mathematics.

Ershow,
Abby
G.,
and
Kenneth
P.
Cantor.
1989.
Total
Water
and
Tapwater
Intake
in
the
United
States:

Population
 
Based
Estimates
of
Quantities
and
Sources.
National
Cancer
Institute
Order
#
263
 
Md
 
810264.

Hansen,
M.
H.,
W.
N.
Hurwitz,
and
W.
G.
Madow.
1953.
Sample
Survey
Methods
and
Theory.
Volume
I:
Methods
and
Applications.
John
Wiley
&
Sons,
Inc.:
pp.
399
and
419.

Kennedy,
W.
J.,
and
J.
E.
Gentle.
1980.
Statistical
Computing.
New
York,
New
York:
Marcel
Dekker,

Inc.:
p.
224.

Life
Sciences
Research
Office
(
LSRO),
Federation
of
American
Societies
for
Experimental
Biology.

Prepared
for
the
Interagency
Board
for
Nutrition
Monitoring
and
Related
Research.
Third
Report
on
Nutrition
Monitoring
in
the
United
States:
Volume
1.
U.
S.
Government
Printing
Office,
Washington,
D.
C.

1995.

National
Research
Council
(
NRC).
1986.
Nutrient
Adequacy,
Assessment
Using
Food
Consumption
Surveys.
Washington,
DC:
National
Academy
Press.

Nusser,
S.
M.,
A.
L.
Carriquiry,
K.
W.
Dodd,
and
W.
A.
Fuller.
1996.
"
A
Semiparametric
Transformation
Approach
to
Estimating
Usual
Intake
Distributions."
Journal
of
the
American
Statistical
Association
91:
1440
 
9.

Recommended
Dietary
Allowances.
1989.
10th
ed.,
National
Academy
Press.
Washington,
D.
C.

Shah,
B.
V.,
et
al.
April
1996.
Statistical
Methods
and
Mathematical
Algorithms
Used
in
SUDAAN
Release
7.0.
Research
Triangle
Institute:
pp.
2
 
5.

Swan,
P.
B.
1983.
"
Food
Consumption
by
Individuals
in
the
United
States:
two
major
surveys."
Annual
Review
of
Nutrition
3:
413
 
32.
5­
7
Third
Report
on
Nutrition
Monitoring
in
the
United
States,
Volume
1,
Appendix
III.

USDA/
ARS
(
U.
S.
Department
of
Agriculture,
Agricultural
Research
Service).
1998.
1994
 
1996
Continuing
Survey
of
Food
Intakes
by
Individuals
and
1994
 
1996
Diet
and
Health
Knowledge
Survey.

CD
 
ROM,
accession
number
PB98
 
500457.
[
Available
from
the
National
Technical
Information
Service,

5285
Port
Royal
Road,
Springfield,
VA
22161.
Phone:
(
703)
487
 
4650.]

Woodruff,
R.
S.
1971.
"
A
Simple
Method
for
Approximating
the
Variance
of
a
Complicated
Estimate."

Journal
of
the
American
Statistical
Association
66:
411
 
4.
