Lead at Superfund Sites: Frequent Questions from Risk Assessors on the Integrated Exposure Uptake Biokinetic (IEUBK) Model
The following frequent questions on the IEUBK model have been divided into four categories:
- Does the IEUBK model address lead in ground water?
- What is a time step and how does it affect IEUBK model predictions?
- Does the variable describing the half maximum saturation of lead apply to the absorbed dose of lead or the total lead intake?
- Does the saturable component only make a difference at intakes greater than 100 micrograms per day (µg/day)?
- What is the appropriate role for a blood lead study?
- EPA guidance on "Calculating Upper Confidence Limits" says that the TRW guidance recommends using average concentrations instead of upper confidence limits (UCLs) (footnote on page two). Where specifically is this recommendation found?
- TRW Recommendations Regarding Gardening and Reducing Exposure to Lead Contamination in Soil
Yes, the IEUBK model addresses lead in ground water.
“Time step” is a variable that determines what averaging time is used to define average daily intakes of lead. For example, if the default time step is considered to be one year (because point estimates may be specified for each year of the seven years of exposure), choosing a time step of one month would result in dividing the lead intake by 12. However, integrating the monthly exposures over a one-year period results in essentially the same lead intake as a one-year time step. For this reason, the model predictions of blood lead concentrations for each age group are essentially independent of the choice of the modeling time step.
The IEUBK model first quantifies the fraction of each media-specific intake that is bioaccessible or available for absorption in the gut. Total intake is calculated as a function of two processes described as saturable and nonsaturable (passive and active uptake in the IEUBK Technical Support Document). The quantity of lead absorbed by the saturable pathway is a function of the total lead available in the gut, the fraction assumed to be absorbed via facilitated passive (PAF) diffusion, and the SATINTAKE variable, which is age specific. Thus, the SATINTAKE variable applies to the intake that is available for absorption.
The non-linearity observed in the intake/uptake relationship from the IEUBK model may be more obvious at higher intake rates (i.e., 100 µg/day or more), but this is only because the equation for a rectangular hyperbola defines the observed curve. At low doses, the relationship is still non-linear, but it is not as obvious.
Blood lead studies need to be undertaken in the context of EPA's policy goals for the protection of public health. In particular, with regard to establishing environmental cleanup levels at contaminated sites, EPA’s Office of Solid Waste Emergency Response (OSWER) has established a health protection goal that young children exposed to lead at their residences should not encounter a risk of more than five per cent of exceeding a blood lead level of 10 micrograms per deciliter (µg/dL). For more information, see the Revised Interim Soil Lead (Pb) Guidance for CERCLA Sites and RCRA Corrective Action Facilities (August 1994) and the Clarification to the 1994 Revised Interim Soil Lead (Pb) Guidance for CERCLA Sites and RCRA Corrective Action Facilities (August 1998). Both are available for download from the Guidance page.
Achievement of this goal depends on an adequate characterization of the lead sources present in a community. Thus, in responding to sites with environmental contamination, EPA is concerned about not only the aggregate community-wide risk, but also the risks to individuals (and particularly children) living at residential locations with high levels of lead contamination.
In the past, blood lead studies have been used for model evaluation and validation. At this time, blood lead studies are more appropriately used for public health monitoring and identifying children at risk.
Blood lead studies address current conditions and do not provide information about future risks
EPA guidance discusses the role and limitations of blood lead studies in addressing future risks: "...Well-designed blood lead studies may be used to identify site-specific factors and pathways to be considered in applying the IEUBK model at residential lead sites. However, OSWER recommends that blood-lead studies not be used to determine future long-term risk where exposure conditions are expected to change over time; rather, they should be considered a snapshot of ongoing exposure under a specific set of circumstances (including community awareness and education) at a specific time. Long-term studies maybe helpful in understanding exposure trends within a community and evaluating the effectiveness of cleanup strategies over time."
Consider the impact of community awareness of lead hazards on observed blood lead levels
Some communities will have received considerable information about environmental lead and potential lead risks to children. It is important to recognize that both direct outreach programs as well as general heightened community awareness have the potential to influence parents’ household hygiene practices and supervision of their children’s activities. At least on a temporary basis, these factors may reduce blood lead levels for some children.
The full range of circumstances that may have served to increase community awareness about lead risks needs to be described and considered in interpreting blood lead study results. While community awareness of lead risks may have served to reduce current lead risks, parental actions to reduce lead contact in the future will be dependent on parents' perceptions about the continuing risks of lead exposure. If parents perceive that there are no longer lead exposure concerns in their community, then motivation for continued efforts to reduce children's lead exposures would likely be reduced. If parents’ awareness of potential risk decreases and the behavioral modifications cease, then higher levels of lead exposure may occur in the future, resulting in elevated blood lead levels.
Recognize complexities in statistical modeling that may be planned
Statistical models are often valuable in the interpretation of environmental and health data. With regard to lead, past modeling efforts have applied regression equations and/or structural equations models to evaluate the dependence of blood lead on environmental lead levels. These tools may have utility in the evaluation of the results from community blood lead studies; however, properties and limitations of these models require specific attention.
In particular, "noise" or "measurement error" in environmental lead levels can lead to biases in statistical modeling. Standard statistical theory assumes that the values of the "independent variables" (here the environmental lead levels) in regression models are known without error. When this is not true and there is substantial "noise" in the measurements, it is widely recognized that regression models can substantially underestimate the true relationships. In practice, there is generally substantial "error" in measurements of soil and dust lead levels. The meaning of "error" in this context is the inability of measured levels to represent children's usual exposures to environmental lead. To some extent, this error can arise from lack of precision in chemical analyses or may be revealed by split or side-by-side samples of environmental media.
However, a larger source of measurement error is generally revealed in variation of soil samples from place to place within a yard (or other sampling unit) and the variability of household dust levels from location to location or week to week. Blood lead studies that are intended to support statistical modeling of the effects of exposure to environmental lead need to place emphasis on evaluating the magnitude of these measurement errors in residential lead levels. This will generally entail collection of repeated independent environmental samples of lead levels at all or a sufficiently large subset of residences. The statistical analysis plan for the study then needs to address the use of collected data on measurement variability in the development of unbiased statistical models.
Another methodological concern is the importance of applying biologically plausible mathematical models to describe relationships between environmental lead and blood lead data. Specifically, if log-scale regression models are fit (relating the logs of blood lead to the logs of environmental lead), a multiplicative relationship is implied between the effects of different environmental media on blood lead. Additionally, a log-scale model will generally imply a strong non-linear relationship between environmental lead and blood lead, even at low levels of exposure. For these reasons, log-scale models have significant limitations for interpreting blood lead data.
Given the complexity of the issues summarized here, expert statistical assistance should be obtained in the development and interpretation of regression or structural equations models fit to blood lead data. Plans for the statistical analysis of study data should specifically address how the plausibility of the mathematical form of any modeling equations and measurement error problems will be addressed. The analysis plan should also provide for the development of error bounds on statistical relationships as well as the provision of "best estimates."
It should also be recognized that statistical models that are fit to blood lead survey data are indeed models and may not necessarily provide accurate representations of empirical truth. The validity of statistical models should be evaluated with a level of scrutiny comparable to that accorded to mechanistic models.
Site-specific calibration of the IEUBK model using blood lead data
At times, it may appear advantageous to plan a blood lead study with an aim to calibrate the parameters of the IEUBK model to match blood lead observations at a particular site. However, in the absence of data providing appropriate site-specific values for particular model parameters, such as soil ingestion or bioavailability, the Technical Review Workgroup for Metals and Asbestos (TRW) cautions against arbitrary adjustments to model parameters to achieve a fit to blood lead data. Such an approach would correspond to simply fitting a curve to the observed data and could effectively circumvent the scientific foundations of the lead model. It should be noted that it may be possible to make a variety of different or incompatible changes to model parameters to achieve a "fit" of model predictions to observed blood lead values. To the extent that data from a blood lead study are judged to be sound and interpretable, these data should form a part of the information presented in the site risk assessment in conjunction with other site data. However, even in the case where modeled and observed blood values are not in accord, an assessment that presents and discusses both types of information can support informed site decisions.
Appropriate uses of blood lead study results
Blood lead data can also indicate the need for further examination of the exposure patterns of children at a site. If comparisons between model predictions and blood lead observations show substantial differences, the reasons for the lack of agreement should be sought. The TRW recommends additional scrutiny of factors that might influence the levels of lead exposures experienced at such a site. Data for households where high blood lead values were seen but not predicted may reveal some personal sources of exposure such as parental occupational exposures. Additionally, if flaking lead-based paint were present, that would reveal a potential for a direct paint chip ingestion (a pathway not included in the model). Additionally, some children may live on relatively clean lots, but be near more contaminated areas. A lack of concordance can also alert site assessors to the possibility that some model parameters (e.g., dust/soil lead ratios, soil ingestion rates or bioavailability of lead) warrant additional direct study.
Avoid unintended effects of interventional efforts in measuring baseline blood lead levels
Cross-sectional blood lead studies need to avoid designs that may inadvertently incorporate interventional features and thus influence blood lead levels they seek to measure. A principal concern regards investigations in which significant contacts with study participants occur at times in advance of the collection of blood lead samples. For example, if residential lead sampling, briefings on study purposes, interviews, or consent forms are completed prior to blood lead sampling, these activities may lead parents to alter practices regarding household cleaning or supervision of children. These actions could subsequently lead to changes in their children's blood lead levels. Changes in parental behavior may result from direct information (e.g., being told that a lead risk is present in their community), implicit information (e.g., being asked to answer a series of questions about their child's behavior or their household cleaning practices), or by inference (from the knowledge that their home and child will soon be screened by expert personnel). There is evidence that individual contacts with parents can contribute to the success of intervention efforts seeking to reduce children's lead exposures. Accordingly, a blood lead investigation that includes significant contact with parents in advance of blood lead collections may implicitly result in unintended interventional effects on blood lead levels.
A period of several months or more is required before children reach a new steady blood lead level after a change in lead exposure. However, as an early phase of readjustment to a change in lead exposures, relatively rapid initial changes in blood lead levels are expected to occur. For this reason, both blood lead and environmental lead samples should be collected at the earliest possible opportunity after the initial contact of study personnel with participants.
Table 1. Precision of estimates for the prevalence of elevated blood lead levels
|Estimated prevalence of blood leads levels >10 µg/dL
|95% confidence interval on prevalence|
|Lower limit||Upper limit|
|Estimated prevalence of blood leads levels >10 µg/dL
|95% confidence interval on prevalence|
|Lower limit||Upper limit|
|Estimated prevalence of blood lead levels >10 µg/dL
|95% confidence interval on prevalence|
|Lower limit||Upper limit|
aCalculations for 0% prevalence added here.
Source: Screening Young Children for Lead Poisoning: Guidance for State and Local Public Health Officials, Appendix B6. Centers for Disease Control and Prevention (CDC), November 1997.
EPA guidance on "Calculating Upper Confidence Limits" says that the TRW guidance recommends using average concentrations instead of upper confidence limits (UCLs) (footnote on page two). Where specifically is this recommendation found?
The arithmetic mean should be entered for soil lead concentration data in the IEUBK model. From the IEUBK User's Guide (section 2.2.4): "The TRW recommends that the soil contribution to dust lead be evaluated by comparing the average or arithmetic mean of soil lead concentrations from a representative area in the child's yard."
The IEUBK model can use an upper confidence limit (UCL); however, the interpretation for the model results is somewhat different if a UCL is used. If an arithmetic mean (or average) is used, the model provides a central point estimate for risk of an elevated blood lead level. If a UCL is used, the model result could be interpreted as a more conservative estimate of the risk of an elevated blood lead level.
For most common garden vegetables, the uptake of metals is not very high. For the most part, exposures would tend to come from consuming adhered soil on unwashed produce (i.e., fruits, such as tomatoes, would be less of a problem than roots or tubers, although these are frequently scrubbed or peeled before consumption). Nevertheless, EPA generally cautions against gardening in areas of known contamination. Also, it may be advisable to NOT consume produce from a garden in the drip line of a home or building structure or from areas where contamination is known to be located.
Another source of exposure related to gardening is handling/intensive contact with contaminated soil and the potential for tracking the contaminated soil into the house (on tools, shoes, or clothing). Vegetables, hands, clothing, and tools should be cleaned before being brought indoors to reduce tracking contaminated soil into the residence.
While 400 parts per million (ppm) lead in soil is generally considered an appropriate screening level for soil lead (unrestricted residential contact to soils where the bioavailability is not greater than default assumption), EPA recommends building raised beds with clean topsoil (no greater than 50 ppm lead) for gardening. In addition, some communities offer free or low-cost plots of clean soil for urban vegetable gardening (community gardens). These recommendations address concerns with track-in of contaminated soil and possible consumption of unwashed produce.
In addition, the USDA has prepared a fact sheet that discusses health risks of lead in garden soil.
While the TRW Lead Committee does not maintain a database of reported uptake rates in fruits and vegetables, several studies provide additional information on homegrown produce and exposure to metals in soil.
- Bechtel, 1998, Empirical Models for the Uptake of Inorganic Chemicals by Plants, BJC/OR-133, Prepared for the U.S. Department of Energy, Office of Environmental Management by Bechtel Jacobs Company LLC, September 1998.
- Boon DY, Soltanpour PN (1992) Lead, cadmium, and zinc contamination of Aspen garden soils and vegetation. J. Environ. Qual. 21: 82-86.
- Corey, J.C., Boni, A.L., Watts, J.R., Adriano, D.C., McLeod, K.W. & Pinder, J.E.d. (1983). The relative importance of uptake and surface adherence in determining the radionuclide contents of subterranean crops. Health Phys, 44: 19-28.
- Davies BE (1978) Plant-available lead and other metals in British garden soils. Sci. Total Environ. 9: 243-262.
- Gzyl, J. (1990). Lead and cadmium contamination of soil and vegetables in the Upper Silesia region of Poland. Sci Total Environ, 96: 199-209.
- Hooda, P.S. & Alloway, B.J. (1996). The effect of liming on heavy metal concentrations in wheat, carrots and spinach grown on previously sludge-applied soils. Journal of Agricultural Science, 127: 289-294.
- Peryea F. (2001). Gardening on Lead- and Arsenic-Contaminated Soils. EB1884. Washington State University Cooperative Extension.
- Pip E (1991) Cadmium, copper and lead in soils and garden produce near a metal smelter at Flin Flon. Manitoba. Bull. Environ. Contam. Toxicol. 46: 790-796.
- Preer JR, Sekhon HS (1980) Factors affecting heavy metal content of garden vegetables. Environ. Pollut. Ser. B. 1: 95-104.
The November 2017 OLEM Directive 9200.2-1, “Recommendations for Default Age Range in the IEUBK Model”, recommends the 12-72 month age range for lead risk assessment at Superfund sites. The IEUBK model software retains the flexibility to run the full 0-84 month age range for which the IEUBK model was validated, because the TRW Lead Committee recognizes that there may be users who require this information. For example, other program offices and researchers who may be interested in the full 7 – year age range also use the IEUBK model. Also, there may be site-specific exposures where older children (for example, 72-84 month trespassers) are considered.
OLEM Directive 9200.2-1 and other directives are available at:
- Does the ALTERNATE SOURCE variable allow the input of different bioavailability values for intake?
- If the soil concentration is changed, does the dust concentration also have to be changed or should it be left as the default unless site-specific data are available?
- Does the IEUBK model allow for the modeling of formula-fed infants ages zero to 12 months assuming the formula is reconstituted with drinking water?
- How much formula does the IEUBK model assume is ingested (in volume/day)?
- How can I derive a site-specific geometric standard deviation (GSD)?
- What should I know before attempting to change the default soil ingestion rate?
The ALTERNATE SOURCE variable allows the user to change from the default value (0 percent) to 50 percent by selecting YES for CHANGE GI VALUES/BIOAVAILABILITY. After selecting YES, the user is notified about changing default values only with adequate and defensible site-specific data. The user is also advised to consult the Technical Support Document for the Integrated Exposure Uptake Biokinetic Model for Lead in Children [NTIS #PB94-963505, EPA 9285.7-22] (December 1994), available on the Software and Users' Manuals page, for a description of bioavailability. The user should select the ESCAPE key (Esc) to access the next screen, where changes to the bioavailability of the alternate source variable can be made. Select ESCAPE again to return to the initial screen. A HELP screen is available to further assist the user.
By selecting the multiple source analysis option, the IEUBK model would automatically calculate the corresponding dust concentration from the soil concentration.
Diet items include:
- canned vegetables and fruits
- fresh vegetables and fruits
- infant items (non-formula items)
Rather than calculating intakes from ingestion rates and concentrations, lead exposure from each food item including formula is collectively estimated by the total dietary lead intake estimates for each one-year age group.
Drinking water is defined as the portion of total water intake that is consumed as direct tap water (see EPA’s 1997 Exposure Factors Handbook). Tap water may be ingested directly as a beverage, or indirectly as an additive to prepared foods. The IEUBK model accounts for lead intake from direct tap water consumption via the Drinking Water menu, and lead intake from indirect tap water consumption via the Diet menu. According to Ershow and Cantor (1989), tap water ingestion rates in liters per day (L/day) for infants may include indirect tap water used in baby formula.
This is complicated by the fact that three types of formulas were reported in the national surveys:
- (1) Ready-to-feed formula – assumed to contain 100 percent intrinsic (i.e., not from tap) water and 0 percent indirect tap water.
- (2) Formula prepared from diluted concentrate (1:1/volume:volume) – assumed to contain by weight 43 percent intrinsic and 57 percent indirect tap water.
- (3) Formula prepared from reconstituted powder – assumed to contain 100 percent indirect tap water.
Reference: Ershow, A.B., and K.P Cantor. 1989. Total water and tapwater intake in the United States: Population-based estimates of quantities and sources. Bethesda, MD: Life Sciences Research Office, Federation of American Societies for Experimental Biology.
The IEUBK model assigns lead intakes in micrograms of lead per day (µg Pb/day) for food items, rather than calculating intakes from ingestion rates (volume or mass/day) and concentrations. The default lead uptake (from direct tap water consumption) for zero to 12 months of age is 0.37 micrograms per day (µg/day), which is based on lead intake from water of 0.80 µg/day (0.20 L/day x 4.0 µg/L). Assuming a 50 percent absorption factor, uptake is approximately 0.40 (the actual uptake is slightly less due to non-linear uptake approach). Using 2.0 L/day instead of 0.20 L/day would yield an approximately 10-fold higher intake from water but this is likely to greatly overestimate direct tap water ingestion, as interpreted by EPA for the IEUBK model.
In general, the TRW does not recommend attempting site-specific GSD estimates. This parameter is particularly difficult to evaluate at a site, as it is demanding with regard to the amount and quality of the data and the potential complications in the analysis. Unless there are substantial differences in child behavior and lead biokinetics at your site, the default GSD should be used (since it is based on national averages). Thus, site-specific GSD values should not be needed (see the Guidance Manual for the IEUBK, EPA 1994, page 4-25, Section 4.2.2, available for download from the Guidance page). In particular, the TRW recommends not substituting site-specific GSD estimates for the default value without detailed, scientifically defensible studies documenting site-specific differences in child behavior or lead biokinetics (see the IEUBKwin User's Guide, page 38, Section 2.2.8, available for download from the Software and Users' Manuals page).
If the blood and environmental lead studies are to be useful for the lead risk assessment, priority needs to be placed on complete reporting of the study. Adequate written documentation would include final study protocols, completed quality assurance documentation (for blood and environmental lead measurements), and adequately reviewed written reports of study findings. The study report should fully address:
- Subject selection procedures and participation rates.
- Timing of all contacts by investigators with study participants.
- Background information on lead provided to the study cohort prior to and during recruitment, sampling or other activities.
- Information made available to the community about lead and this study prior to and during initiation.
- Results of the environmental sampling and associated quality assurance.
- Results of blood lead sampling and associated quality assurance.
- Pertinent demographic and behavioral data collected during the study.
One component of the final study report should be a paired data set (by child and household) with blood lead and environmental lead measurements, sampling dates, and pertinent behavioral and demographic information (including time spent in day care and other settings away from the home).
The TRW strongly recommends writing up and reviewing any community blood lead studies independent of other site documentation (i.e., separate from the site risk assessment). The site risk assessment would certainly draw on the results of the blood lead investigation; however, the reporting of the blood lead study should be adequate to stand alone. Hopefully, this approach may also allow a rapid completion of the blood lead study and aid in site decision-making.
Recognizing the technical difficulties of interpreting soil and dust ingestion studies, the 1995 Administrative Reform for lead specified that adjustments to the IEUBK model default ingestion rates only be performed following a recommendation from the Office of Emergency and Remedial Response (EPA-OERR, now the Office of Superfund Remediation and Technology Innovation [OSRTI]) that such a change is appropriate. The short sheet IEUBK Model Soil/Dust Ingestion Rates discusses the earlier review of soil ingestion studies to determine if adjustments to the soil ingestion rates used in the IEUBK model are warranted.
- Short Sheet: IEUBK Model Soil/Dust Ingestion Rates (PDF)
In 2006, the TRW reviewed the soil ingestion literature published since the short sheet was written to assess whether data from recent studies indicated changes to the soil ingestion rates were warranted. Based on its review, the TRW concluded that the recent literature did not support changes in the default soil ingestion rates. In fact, much of the literature published since 1999 reanalyzed data from studies performed prior to 1999 rather than adding new data to the field.
Likewise, a recent independent review by EPA's National Center for Environmental Assessment in the Child-Specific Exposure Factors Handbook did not support any changes to the recommended estimate of the mean soil ingestion rate for children.
- Child-Specific Exposure Factors Handbook
- How many significant digits are read from a batch mode input file?
- Why is the predicted blood lead concentration (PbB) value 1.15 micrograms of lead per deciliter of blood (µg/dL) when input values are all zeros?
- From a list of input files, can the IEUBK model automatically read the next source file?
- How can batch mode be used to calculate future risk?
The IEUBK model considers only two significant digits from the data input file for running batch mode. However, the model uses double precision (eight digits) in all its calculations.
The PbB value is 1.15 µg/dL because in batch mode the contribution from other dietary sources is always present.
Yes, this feature is now available.
For future scenarios, or in the absence of age data for each residence, the TRW recommends that the 32-month age group be used in batch mode runs. This age result approximates the 12- to 72-month average calculated in single run mode. Although some slight differences result between the geometric mean blood lead concentration (PbB) and P10 value for the 32-month age group compared to the 12- to 72-month average, the differences in the results are so small that they are not expected to affect site decisions.
- Can the recommended holding times be exceeded for analysis of lead in soil samples?
- How should garden vegetable samples be collected for use in the IEUBK model?
- At what depth should soil samples be collected from for risk assessment purposes?
- What sampling depth is most representative of surface soil and dust that is associated with exposure (both direct contact and incidental ingestion) of children? Is this sampling depth also representative of the dirt that is applicable to the mass fraction of soil in indoor dust (MSD) term?
- Are there any interior dust cleanup confirmation sampling protocols that should be used in addition to the Toxic Substances Control Act (TSCA) wipe sampling protocols?
- Can the IEUBK model be used to develop a site-specific fish advisory?
- Answers to Frequently-Asked Questions for XRF (x-ray fluoresence)
Current EPA recommendations are that holding times for metals analysis of soils should not exceed six months. However, in its 2005 Sample Holding Time Reevaluation (PDF)(329 pp, 2 MB) study, EPA evaluated sample holding times for metals analysis of soils and found that for a holding time of one year, no chemically significant change in concentration occurred. This study also found no significant difference between moist or dry sample handling. In addition, NIST and ASTM as well as other certified soil standards for metals have typical life expectancies of up to 10 years without significant changes in metal concentration under a variety of storage conditions.
Based on the soil standards and the 2005 EPA study, the TRW recommends that holding times for soil lead determination may be extended up to two years total and that refrigeration is not necessary for total soil lead (though a site-specific QAPP may specify a different holding time or storage conditions) with reasonable expectation that representative results will be obtained.
The alternate dietary values feature of the IEUBK model enables risk assessors to predict impacts of ingestion of lead contained in locally harvested foods (e.g., fruits, vegetables and game) on blood lead concentrations of children. This feature can be used when site-specific data are available to estimate both the concentration of lead in these food sources and the contributions that these food sources make to the diet of a typical child at the site.
Lead exposure concentration: The exposure concentration term should represent the concentration to which children are exposed; that is, it should be an estimate of the concentration of lead in the food as prepared for ingestion. To ensure that the measurements of lead in homegrown vegetables reflect the expected exposure concentrations as closely as possible, measurements should be made on vegetables that have been prepared for ingestion. In addition, in deriving the concentration term and interpreting the results of IEUBK model runs, consideration should be given to the potential pathways by which lead may enter or be associated with homegrown vegetables.
These pathways include: 1) incorporation of lead into vegetable tissue during growth; 2) deposition of lead on the vegetables during growth or harvesting (e.g., soil-derived dust); and 3) deposition of lead on the vegetables during processing, preparation or storage (e.g., paint- or soil-derived dust). The extent to which any or all of the above pathways will be represented in measurements of lead in garden vegetables will depend on the sampling and analytical designs. Measurements made on unwashed garden samples may reflect lead deposited on the plants that might not be ingested after typical preparation of the vegetables for meals. If these data were used as alternate dietary values in the IEUBK model, the impacts of garden vegetable lead on blood lead concentrations may be overestimated. On the other hand, measurements made on washed vegetables, that might be consumed by children without washing (e.g., tomatoes), may underestimate blood lead impacts if these data were used in the model.
The above considerations emphasize the importance of establishing an exposure pathway model before designing a sampling approach that will provide adequate data to support the exposure model. This model should consider the types of vegetables most likely to be consumed by children, the pathways for lead incorporation or association of lead with those vegetables, and the ways in which the vegetables are likely to be prepared (or not prepared) for consumption by children. In addition, several potential uncertainties should also be considered in interpreting model output:
- How much of the lead measured in the vegetables represents soil-derived or other sources of dust, that may already be accounted for in the estimate of soil and dust lead intake?
- Do the lead concentrations measured in the vegetables actually reflect the lead concentrations in the vegetables consumed by the children? That is, do the children actually eat the vegetables included in the vegetable lead survey? If lead estimates for various vegetable types are averaged or aggregate samples are collected, does the distribution of the aggregate accurately reflect the distribution consumed by children? Also, is the lead measured in the vegetables actually consumed? For example, where is the lead in vegetable soup? In the stock, or the vegetables? Which was actually measured in the lead assay?
- Sampling bias and measurement error.
Although it may not be possible to remove these uncertainties entirely from the model, sensitivity analyses can be used to set plausible bounds on the potential impacts of the homegrown vegetable pathway on blood lead concentration.
Lead absorption: The default absorption fraction of 50 percent for lead in food at low intakes is applied to all dietary intakes, including homegrown vegetables. Unless there are data supporting this value, it may be appropriate to reduce this value. The appropriate value will depend on the source of the lead associated with the vegetables. For example, if most of the lead measured in vegetables is derived from contamination with soil-derived dust during harvesting, then the soil lead absorption factor (ABSS) of 30 percent may be more applicable. The use of a value of 50 percent for the absorption factor may result in an overestimate of the uptake of lead from ingestion of homegrown vegetables. It may be appropriate to make this a user-selectable option and explain the options in the guidance and on the help screen.
It is recommended that sampling designs be developed to provide the necessary data for all phases of a cleanup project (e.g., human and ecological risk assessment and remedial design) within one sampling effort to minimize mobilizations whenever feasible. This frequent question response provides recommendations for sampling depth for risk assessment, where the primary objective of the sampling effort is to estimate an average soil lead concentration for use in the IEUBK model. Recommendations in this FAQ response should be incorporated within the sampling design developed for the site. This frequent question response assumes that data on the extent (e.g., depth) of contamination are already available for the site (e.g., from the Site Assessment process) or will be provided pursuant to other objectives of the sampling design.
The appropriate sampling depth depends upon the conceptual site model (CSM) and the exposure scenario for the site. There may be more than one exposure scenario for the site, and therefore more than one CSM. For example, one exposure scenario on a site may be children playing in a residential yard with exposure to contaminated surface soil. The same site might also include a plausible scenario that involves the exposure of residents or construction workers to subsurface contamination (e.g., septic system repair, gardening; see the Superfund Lead-Contaminated Residential Sites Handbook, available on the Guidance page). The sampling depth should be appropriate for the exposure pathways and contaminant transport routes of concern, and should be chosen with these considerations in mind.
Keeping in mind the broader considerations above, to assess risk from current exposure to contaminated surface soils, EPA has recommended the collection of surface soil from the top two to three centimeters (zero to one inch) of the soil layer, below organic litter or sod. For more information, see the 1996 EPA Soil Screening Guidance, available on EPA’s Superfund Soil Screening Guidance Web page. The TRW and the LSW agree that this depth best represents the soil and dust exposure for predicting child blood lead level using the IEUBK model, as well as for estimating the IEUBK's mass fraction of soil to dust parameter (MSD). For example, see the Superfund Lead-Contaminated Residential Sites Handbook (2003), available on the Guidance page. See also the TRW Recommendations for Performing Human Health Risk Analysis on Small Arms Shooting Ranges (2003), available on the Guidance page.
These guidance documents recommend sampling from the top two to three centimeters (or shallowest depth that can be reasonably obtained, see below) because children are typically exposed to surface soil. These recommendations were intended to avoid using data from samples collected at depth (e.g., 0- to 6-inch depth interval) that might dilute contamination that is concentrated in the surface soils, thereby underestimating the exposure (and therefore risk) to children. If the concentration of lead is relatively homogeneous across the vertical extent of contamination, the potential for dilution does not exist; therefore, it makes no difference what depth interval the samples are collected from, provided they are collected from within the zone of relatively homogeneous contamination. If contamination is found in subsurface soils (i.e., greater than zero to one inch below the ground surface), then the risk assessment for the current exposure scenario should consider the likelihood that children may be exposed to soils at that depth, and select the sampling depth accordingly.
Samples collected at depths greater than one inch below the ground surface may also be appropriate for future use scenarios (e.g. gardening, construction activities, yard maintenance). To assess risks from exposure to contaminated subsurface soils, samples should be collected from the depth interval consistent with the applicable exposure scenario. Samples below one inch are also useful for determining where institutional controls may be needed; contamination at depth that is left in place as part of the remedial action warrants institutional controls, For more information, see the Superfund Lead-Contaminated Residential Sites Handbook (2003), available on the Guidance page).
Sampling depth also varies depending on site-specific conditions. The Risk Assessment Guidance for Superfund (RAGS) Part A (EPA, 1989) states that the assessment of surface exposures will be more certain if samples are collected from the shallowest depth that can be practically obtained. At some sites, it might be possible to collect a sufficient quantity of soil at depths less than two centimeters (e.g. 0-to-1-centimeter depth interval). At other sites, it may be difficult to obtain the required amount of soil material from the top two centimeters (e.g., due to rocks or debris). In these instances, the required quantity of sampled material should be obtained by slightly increasing the area sampled, rather than increasing the depth of the sample, to avoid the potential for diluting surface soil contamination (see above).
Finally, the exposure point concentration for each exposure scenario should be estimated with data from the depth interval(s) relevant to each scenario.
What sampling depth is most representative of surface soil and dust that is associated with exposure (both direct contact and incidental ingestion) of children? Is this sampling depth also representative of the dirt that is applicable to the mass fraction of soil in indoor dust (MSD) term?
EPA has recommended the collection of surface soil from the top two centimeters (zero to one inch) of the soil layer for use in baseline risk assessments (see the 1996 EPA Soil Screening Guidance, available on EPA’s Superfund Soil Screening Guidance Web page). The TRW agrees that this depth best represents the soil and dust exposure for use in calculation of the predicted child blood lead level using the IEUBK model as well as characterization of the MSD.
Both the concentration of lead in dust and the loading of accessible dust in a home will affect the current risks to a resident child. For a full picture of risk conditions in an individual home, both types of data would be recommended. It is important to note, however, that wipe samples that address only lead loading do not allow an understanding of whether lead that is present is in the form of a small amount of high-lead-concentration dust material or a larger amount of lower-concentration material. As such, wipe samples do not provide as much information as appropriate vacuum samples from which both concentration and loading may be measured.
Considerations for sampling conducted to judge the immediate effectiveness of a dust-cleaning action will be different from considerations for sampling to develop information that can be used in risk assessment. Lead measurements taken immediately after a cleaning operation can reflect the effectiveness of that cleaning (in comparison with prior measurements), but will reflect too transient a set of conditions to estimate continuing risks a child living in the house may encounter.
It is important to recognize that dust remediation does not remove lead from the dust – it removes the dust. So, immediately after a cleanup action, one might have difficulty collecting an adequate size dust sample, and it is likely to have a similar concentration to the pre-remediated dust. In the days or weeks following a cleaning, dust loadings (although not necessarily lead loadings) would be expected to increase, depending on the transport of dust materials into a home, the source of these materials and redistribution of dust within the home. Thus, samples used to predict the continued risks of children would best be collected after some amount of time has passed following a cleaning action to allow for equilibration of dust lead concentration with sources outside the home (e.g., one month or more). EPA's validated risk methodologies for lead (using the IEUBK model) require an estimate of dust concentration as an input. Lead concentration is also a good prospective indicator of risk, as it should be less subject than dust loadings to sharp changes that are expected with each normal cleaning and vacuuming cycle in a home.
A sampling method that provides loading data is appropriate for clearance confirmation; however, it does not matter whether wiping or dry-vacuuming is used, as long as a loading value can be obtained. The goal of sampling would be to demonstrate either a significant reduction (assuming that pre-remediation data are available) or achievement of a clearance standard such as the TSCA 403 standard.1
Although it is more difficult and expensive, a vacuum sampling device with a template to define the sampling area may be used to simultaneously collect: 1) lead loading; 2) dust loading; and 3) lead concentration data. This alternative is valuable if you want to use the data for both clearance and risk assessment.
The studies listed below suggest that as predictors of blood lead concentration (PbB), the results are comparable between vacuuming and wipe samples. One limitation is that these methods do not explicitly measure the rates of lead and dust deposition, unless you conduct repeat sampling.
Lanphear, B.P., Emond, M., Jacobs, D.E., Weitzman, M., Tanner, M., Winter, N.L., Yakir, B. & S. Eberly. 1995. A side-by-side comparison of dust collection methods for sampling lead-contaminated house dust. Environ Res. 68: 114-123.
Rust, S.W., Burgoon, D.A., Lanphear, B.P. & S. Eberly. 1997. Log-additive versus log-linear analysis of lead-contaminated house dust and children's blood-lead levels. Implications for residential dust-lead standards. Environ Res. 72: 173-184.
Emond, M.J., Lanphear, B.P., Watts, A. & S. Eberly. 1997. Measurement error and its impact on the estimated relationship between dust lead and children's blood lead: Members of the Rochester Lead-in-Dust Study Group. Environ Res. 72: 82-92.
1 According to the TSCA 403 Rule, a dust-lead hazard is surface dust in a residential dwelling or child-occupied facility that contains a mass-per-area concentration of lead equal to or exceeding 40 mg/ft2 on floors or 250 mg/ft2 on interior windowsills based on wipe samples.
Yes. In evaluating the fish consumption limits for a specific site, the IEUBK model may be used. The dietary lead input window of the IEUBK model has an option to use alternate dietary lead intake. The alternate dietary intake menu includes ingestion of game animals from hunting, fish from fishing, and homegrown fruits and vegetables. The IEUBK model uses the total lead uptake calculated from the estimated intakes of lead in all media (soil, dust, air, water, and diet) to predict mean blood lead concentration for children (84 months or younger).
The IEUBK model assumes the substitution of fish for other meat dishes, so an estimate of fish meals as a proportion of total meat meals is required for the exposure calculation. In estimating lead exposure to children from recreational fishing, appropriate inputs for percent of food class would be 10 percent for recreational fishing or as much as 50 percent for the high-end case. The average concentration of lead in fish as consumed (either filets or whole fish) is entered as concentration (µg Pb/g of fish tissue).
The IEUBK default total meat intake (grams per day) for different age groups is given as follows:
- zero to 11 months (29.6 g/day)
- 12 to 23 months (87.4 g/day)
- 24 to 35 months (95.7 g/day)
- 36 to 47 months (101.6 g/day)
- 48 to 59 months (107.4 g/day)
- 60 to 71 months (111.9 g/day)
- 72 to 84 months (121.0 g/day)
In addition, EPA Guidance for Assessing Chemical Contaminant Data for use in Fish Advisories Vol. 2 (EPA 823-B-97-009) provides for three-ounce meals (85 g/day) for children; therefore, consumption of three-ounce and an upper bound eight-ounce (85 grams and 227 grams, respectively) fish meals may also be used in the analysis. The soil and dust lead concentrations are important input values for the IEUBK model. A site-specific arithmetic mean soil lead concentration could be used, or one could calculate the 95 percent upper confidence limit (UCL) on the arithmetic mean for the site as a whole considering soil remedial action at the residential site-specific cleanup level. A site-specific dust concentration could be entered or, if these data are not available, then the Multiple Source Analysis relationship may be used.
- Example illustrating an approach one may use to develop a fish advisory using the IEUBK (PDF) (3 pp, 159 K)