S-EPA
United States
Environmental Protection
Agency
Health Effects Research
Laboratory
Cincinnati OH 45268
Research and Development
EPA-600/S1-81-058 Aug 1981
Project Summary
Development of
Methodology for Determining
Risk Assessment When
Sludge is Applied to Land
Peter R. Jutro and Anil Nerode
This project explored the feasibility
of developing a risk assessment
methodology that could be applied to
sludge management decision making.
It examined cadmium, since this sub-
stance is one of the best studied and
most extensively reported contami-
nants.
The methodology developed allows
determination of the proportion of the
population experiencing given levels
of exposure to a toxic substance under
specified management strategies. In
addition, it provides for the evaluation
of the damage caused by such expo-
sure. When both the dosage-response
analysis and the exposure population
analysis are joined, the distribution of
population into levels of response can
be established. Each requires separate
categories of data. The first is the
result of controlled experiments and
carefully designed epidemiological
statistical studies. The second con-
sists of ongoing data bases on a
national scale, including both meas-
urements of background levels of the
toxic substance and data on the actual
levels at the various stages of the dis-
posal and dispersal procedures. This
report establishes that these data
bases are required. The problems
associated with both categories of
data are considered and discussed.
To determine feasibility of risk
assessment, the study began an ex-
ploratory analysis of modeling, data
bases and transfer characteristics
needed for an actual risk assessment.
Existing data bases were reanalyzed
on a single systematic basis in a form
which was suitable for undertaking
the exercise. This study used non-
parametric, robust and resistant sta-
tistics for determining empirical distri-
butions characterizing the desired
transfer characteristics.
On the basis of the study, it appears
feasible to use risk assessment for
decision making on toxic substance
disposal. Much of the required data is
being gathered currently, but in an
inefficient fashion not appropriate for
risk assessment. It will be necessary to
formulate appropriate data gathering
and record keeping protocols. This will
eventually happen as regulators and
industry realize that the only way to
defend their decisions is through such
risk analysis using the best current
data and techniques.
This Project Summary was develop-
ed by EPA's Health Effects Research
Laboratory, Cincinnati, OH. to an-
nounce key findings of the research
project that is fully documented in a
separate report of the same title (see
Project Report ordering information at
back).
-------
Discussion
The purpose of this project was to
determine the feasibility of applying risk
assessment methodology to sludge
management decisions. The results
might be used as a possible model for
the larger problem of environmental risk
assessment. The latter was the subject
of two National Academy of Sciences
reports on chemicals in the environ-
ment, which suggested directions for
risk assessment methodology
The risk assessment problem for
environmental decision making became
more prominent after the appearance of
Lowrance'S Of Acceptable Risk. It was
intended to delineate the distinction be-
tween risk and safety, where risk analy-
sis was a quantitative assessment of
consequences of decisions, and safety
analysis was the assessment of level of
risk acceptable to society. This report
accepts that distinction, and is confined
to risk, not safety.
One approach to risk assessment
evaluates the fundamental quantity in
determining risk to the populace (in the
case of effects of toxic substances and
in some other situations) by the degree
of exposure of the populace, determined
by the summing of all exposures along
alternate pathways by which the toxic
substance reaches the populace.
This approach is refined to give a
methodology which, in principle, would
determine what proportion of the popu-
lation experiences each level of expo-
sure to the toxic substance under some
specified management strategy for
handling the substance. Ideally, this
would consist of charting the move-
ments of the substance from source to
final human exposure along the
principle exposure paths followed by the
substance, allowing summation of all
exposure to obtain total exposure. This
is an ideal aim, but in the presence of
limited empirical data, ingenuity is
required to establish approximate total
exposure.
A second part to risk analysis, assum-
ing that the first approach was empiri-
cally determined, concerns the damage
such an exposure may cause in terms of
societal or individual losses. This is
substantially the question of both:
(1) establishing dose-response rela-
tions, in which the dose is the exposure
level, and (2) measuring the response
either in physiological terms (e g., organ
load of a toxic substance), or in epidem-
iclogical terms (such as excess
morbidity and mortality). It is only when
both exposure population analysis and
dosage response or epidemiological
analysis are joined that the distribution
of population into levels of response can
be established. Safety analysis can then
be undertaken determining whetherthe
population response does or does not
represent acceptable damage. Also,
whether an alternative management
strategy is called for, which reduces the
fraction of the population with damag-
ing response above a certain level.
Two separate categories of data
requirements are involved The first
consists of results of carefully control-
led experiments and carefully designed
epidemiological statistical studies. In
the case of land application of sludge,
such results and others are required for
plant uptake data from cadmium in
sludge, human uptake data from cad-
mium in food, animal uptake data from
cadmium in grain, physiological re-
sponse data for ingestion of cadmium in
food, etc. The second category of data
consists of ongoing data bases which
are on a national scale. These include
both measurements of background
levels of the toxic substances in the
medium involved, and data on the actual
levels persistent throughout the country
at the many stages of the disposal and
dispersal procedures. This is a massive
data base requirement. Many of the
results, however, would be invaluable
for many risk assessment efforts other
than sludge management.
Concerning the first category of data,
one reason that the dosage response
data has not been sufficiently empha-
sized in assessing damage is an
assumption that the response is either
proportional to the total dose received,
or given by an a priori function of dose,
independent of the circumstances. Two
reasons that epidemiological studies of
health effects have not been used
extensively are the possibility of con-
founding causes for observed
phenomena other than those that are
apparent in an uncontrolled "experi-
ment," andthe immense size of the data
bases and processing effort required. It
is important to stress that, very often,
only the relative risks of alternate
disposal plans can be estimated well,
and not the absolute risks of any one.
Thus, one may determine that one route
will lead to many times the exposure of
another, without being able to specify
any one's exposure consequences
accurately.
Since there is interest in simultan-
eously controlling at acceptable levels
all toxic substances in the environment,
it is necessary to keep records for many
substances at once in the national data
bases. This makes the whole problem
multivariate and vector, rather than uni-
variate and scalar. It complicates deci-
sion making, since a strategy which will
lower exposure to one toxic substance
may very well raise exposure to another.
For this reason, no rules for safety based
on risk analysis (e.g., population expos-
ure, dosage-response, epidemiological
response) are likely to be forthcoming.
The principal purpose of such analyses
is to show the probable consequences
in the form of response and damage that
each alternative strategy will entail.
Then comes the problem of deciding
among the competing alternatives.
Each alternative entails resulting popu-
lation exposure and population
response distributions of each of 'n'
toxic substances This results in a vector
of exposure distributions and a vector of
response distributions. When specifi-
cally applied to sludge management
decisions, each proposed strategy for
sludge disposal leads to both a popula-
tion exposure and a population
response distribution for each toxic
substance in sludge.
The disposal alternatives for sludge
are: pyrolysis or incineration, water
disposal (e.g., ocean dumping), and land
application (including both agricultural
use and simple land storage and dis-
posal). Each alternative has well-known
general advantages and disadvantages.
Incinerated sludge yields air pollution
byproducts which impact the environ-
ment, enter the food chain, and produce
health effects directly through lung
respiration. Ocean dumping disturbes
ocean ecosystems, the marine environ-
ment and the sea animal food chain to
man. Land application and land disposal
result in increased plant uptake of sub-
stances which enter the food chain to
man directly through human ingestion
and indirectly through use as animal
fodder. It also causes toxic substances
to leave application areas, in runoff,
entering lakes, rivers and groundwater.
Each alternative yields deleterious
effects to man, due to breathing, eating,
or drinking toxic substances, and each
alternative has its own health effects.
The purpose of risk analysis in sludge
management, therefore, is to assess the
relative effects of toxic substances on
human health in the alternative dis-
posal schemes.
This requires the establishment of {
population exposure distributions for
each toxic substance in sludge for each
-------
disposal alternative. It requires estab-
lishment of dose-response and
epidemiological results to determine
the population response distributions
for each health effect dependent upon
each toxic substance for each alterna-
tive disposal scheme. Since population
exposure depends on some kind of
summing over all pathways leading to
man, and each pathway involves differ-
ent transport mechanisms, different
transformations of form for the contam-
inants, different mgestion routes in man
with differing efficiencies, and different
physiological systems exposed to the
contaminants, this is a very complicated
endeavor
This report is restricted to what can be
obtained from current data bases in
both the literature and in computer
storage systems. It was decided to
examine one important substance on
which a great mass of literature has
currently been assembled, cadmium in
sludge The reasoning is that if risk
assessment is feasible for any contami-
nant in sludge, it should be feasible for
cadmium. This, therefore, is a "best
possible" case for risk assessment. If it
is infeasible for cadmium in sludge, it is
very likely to be infeasible altogether
The only drawback is that although cad-
mium is very harmful on general
chemical grounds in the forms that
might be ingested from sludge, very
little has been done for dosage-
response studies to demonstrate
adverse health effects In addition, the
epidemiological approach for demon-
strating increased mortality, morbidity
and chronic effects has not been sys-
tematically applied.
The best way to determine feasibility
of risk assessment for cadmium in
sludge was to begin an exploratory
analysis of modeling, data bases, and
transfer characteristics needed for an
actual risk assessment. This did not
include evaluation of data bases needed
for all subsystems which enter into
population exposure calculations, or for
the epidemiological approach to deter-
mining health effects
The evaluation of the effects of
sludge, independent of which disposal
alternative or combination of alterna-
tives is evaluated, include sources of
sludge, measurement of cadmium
content as a function of sewage plant
characteristics, and output variability
The risk of any one disposal method
can only be evaluated after establish-
ment of a large reliable data base for
each stage of translocation and trans-
formation of the sludge cadmium. Since
systematic risk evaluation as a subject
is in its infancy, some necessary data
bases do not currently exist. Those that
do exist, having been accumulated for
other purposes in existing measure-
ment programs, may not be adequate to
describe the transfer features from
stage to stage of the disposal process.
What can be done at present is to
reanalyze the existing data bases on a
single systematic basis. This would be
in a form suitable for establishing the
desired transfer characteristic for each
stage and combining them to obtain a
final population exposure analysis. Also
to see if it is feasible to upgrade these
measurement programs and to
introduce new ones to fill in the present
gaps, so as to demonstrate the feasibil-
ity of at least population exposure analy-
sis for each disposal method
This study used non-parametric,
robust and resistant statistics, with the
purpose of determining empirical distri-
bution characterizing the desired trans-
fer characteristics The data bases used
are as follows For land application,
there is information on soil uptake of
cadmium from applied sludge and
applied cadmium salts under a variety of
conditions, and on plant uptake, both as
a function of soil cadmium and as a
function of applied cadmium in sludge
or salt form. Some literature is available
on animal and human food uptake of
cadmium from cadmium-containing
food, and on distribution of cadmium in
human diets (market basket, institu-'
tional diets). There is an extremely
amorphous literature on the health
effects of dietary ingested cadmium
This information was accumulated
neither on a common statistical basis
nor on a transfer characteristics basis It
was necessary to reanalyze all this
literature data from its raw form to
obtain an insight as to whether further
experiments designed to fit into transfer
form for risk assessment would leadtoa
reliable population exposure analysis.
Even the proper indicators (measure-
ments of appropriate observable quanti-
ties) have not been standardized; these
are such quantities as cadmium content
of sludge, available soil cadmium,
standard soils, standard definition of
plant load or exposure, and standard
definition of human load or exposure
through dietary mgestion. This is parti-
ally due to what exact forms the cadmi-
um takes, and the strong association of
their effects with those of other heavy
metals Another problem concerns the
lack of human and animal in vivo non-
destructive testing for which there are
methodologies not yet used at this time.
A non-parametric reanalysis of the
available data in this area was done.
The purpose of these reanalyses was
to determine whether natural refine-
ments of present day experimental
measurement practice would lead to a
coherent set of transfer characteristics.
These would yield, for each relevant
form of cadmium, the proportion of the
population with a given incremental
exposure due to a given regimen of land
application of sludge Failing this, if
enough identification of transfer char-
acteristics would result to identify the
feasibility of controlling certain stages
of the sludge disposal system, the
analysis of the proportion of population
with a given exposure level might result
in a predetermined way at specifiable
cost.
The remaining problem concerns the
determination of the human health
effects at a given exposure level. This
would be the result of clinical trials, of
extrapolation from animal experiments,
or of epidemiological studies of correla-
tion of mgestion levels with likely
mortality and morbidity and chronic
symptoms on a population basis. The
former is preferable to the latter, but
often the epidemiological route is the
only one available in exploratory data
analysis, due to the lack of suitable
clinical experiments. In the present
case, the information on'dietary distri-
bution of cadmium available through
regional market basket surveys is so
inadequate in method and sample, that
correlation with regional health effects
is simply not possible, even on an
exploratory basis. This is due to the
chosen sampling technique. For the
health effects of given exposure level to
dietary cadmium, therefore, there is no
easy solution. Current technology may
be adequate to supply some dose-
response information on an in vivo basis
once the usefulness has been empha-
sized. Past measurements of cadmium
level in cadavers with given symptoms
apparently does not contribute much
information for the transfer character-
istics of dose-response needed for
exposure-health effect analysis for risk
assessment, since the corresponding
exposure levels of the cadavers are
unknown.'
Recommendations
(1) Risk assessment of sludge
management should assess
-------
comparative population exposure
and health effects obtained for
each of the main sludge manage-
ment alternatives: land applica-
tion, incineration, and ocean
dumping
(2) Risk assessment of these alterna-
tives in sludge management is
best carried out in the context of a
general risk assessment program
for heavy metals and probably for
toxic substances. The many nec-
essary data bases for each risk
assessment for sludge manage-
ment are certain to be used as
subsystem descriptions for risk
assessments for other toxic
substance management problems
not involving sludge management.
(3) The empirical distribution of the
nation's sludge output, character-
ized by heavy metal levels, needs
to be collected on an ongoing
basis. A record keeping and
reporting system should be man-
dated. Exploratory data analysis
on preliminary data indicates that
the monitoring resources should
be utilized to maximize the num-
ber of sludge plants sampled,
rather than the number of times a
single plant is sampled.
(4) The distribution of land area by
yearly application rate of sludge
needs to be evaluated by sampling,
together with descriptions of land
type and crop The result is the
cadmium loading pattern Explor-
atory data analysis reveals that
cadmium availability for plant
uptake depends on recency of
application. This possibility should
be investigated as a potential con-
trol point in the land use strategy;
it should certainly be incorporated
as a design element in any future
studies of plant uptake following
sludge application.
(5) For a variety of plants (corn,
soybeans, etc ) the relative propor-
tion that each one accounts for in
the mix of crops that will be grown
on sludge-amended soil should be
ascertained, as should the distri-
bution of cultivars within species
The analysis should be separated
at this point by plant type.
(6) From existing data and from new
work, as required, the response
(uptake and translocation) of these
cultivars and species needs to be
found, or a plausible distribution
of such responses needs to be
estimated, and the resulting distri-
bution convolved with the one
described in recommendation 4,
the cadmium loading pattern. One
may imagine that the problem has
been refined to that proportion of
the whole cadmium problem that
is represented by the fraction of
the total acreage that is devoted to
the particular cultivar.
(7) The distribution should then be
refined further by plant part, and
the task here is simply to keep
track of the separated distribu-
tions of uptake in leaf, grain, and
stover.
(8) Population physiological and
epidemiological response to
dietary cadmium exposure has not
been established by current
research. This is a necessary com-
ponent of risk assessment if the
latter is to be based on population
response, rather than population
exposure
(9) The measurements needed forthe
forms of cadmium at each level of
the risk analysis process have not
been standardized. This is essen-
tial for uniformity and reliability of
results. This is the problem of
validating the indicators of risk.
(10) The data bases are not organized
nationally so as to facilitate popu-
lation exposure analysis, i.e.,
exposure to cadmium via air, food
and water. Thus, programs for
accumulating the missing bases
and programs for rationally stor-
ing the bases in compatible form
need to be developed
(11) It is desirable to collect the
bivariate distributions expressing
the dependencies of pairs of
variables (assuming to be inde-
pendent in the stage diagrams) as
well as the marginals of these
distributions described above, at
least for all heavy metals.
(12) The statistical methodology for
estimating distributions that
result from the combinations used
for population exposure analysis
from sampled data for the com-
ponent distributions should be
developed further The non-para-
metric approach appears to be
best suited mathematically forthis
purpose
-------
Peter R. Jutro and Anil Nerode are with Cornell University, Ithaca, NY 14853
Norman E. Kowal is the EPA Project Officer fsee below).
The complete report, entitled "Development of Methodology for Determining
Risk Assessment When Sludge is Applied to Land, "(Order No. PB81 -240 012;
Cost: $15.50, subject to change)'will be available only frorrr
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at.
Health Effects Research Laboratory
U S. Environmental Protection Agency
Cincinnati, OH 45268
US GOVERNMENT PRINTING OFFICE, 1981 —757-012/7319
------- |