Estimating Exposure Point Concentrations

Learn how to use ATSDR’s guidance to estimate appropriate EPCs to use in your site-specific exposure calculations.

An EPC is a representative contaminant concentration that is calculated for each:

  • Exposure unit or area
  • Individual completed and potential exposure pathway.
  • Exposure duration: acute (0–14 days), intermediate (15–364 days), or chronic (365 or more days).

When appropriate, health assessors can use statistical tools to help calculate EPCs based on the available sampling data. ATSDR has developed very specific guidance for health assessors on how to estimate EPCs for either discrete sampling or non-discrete sampling data. When performing this step in the PHA process, always refer to ATSDR’s Discrete Sampling Guidance for individual environmental samples from a given location and time that are independent of other samples, and ATSDR’s Non-Discrete Sampling Guidance for multi-incremental sampling and composite sampling data. These guidance documents outline how to calculate EPCs for your specific scenarios. You can also contact the ADS group for more information or an ATSDR statistician for questions about EPC estimates.

ATSDR’s Public Health Assessment Training (PHAT) EPC Mini-Module [PDF – 2.2 MB] provides information and training on estimating EPCs for some example scenarios. But your calculations will depend on your specific data set, including the number of samples, amount of non-detect data, the distribution of the data, and whether there are special contaminants to consider, such as polycyclic aromatic hydrocarbons (PAHs), dioxins, and asbestos.

Examples of Questions to Ask When Calculating an EPC
  • Who are the exposed groups?
  • What is the exposure duration?
  • What types of samples were collected (discrete, non-discrete, both)?
  • How many samples were collected in the exposure unit?
  • Are there non-detect data points?
  • How many samples were analyzed for the contaminant?
  • How many samples was the contaminant detected in?
  • What percent of samples was the contaminant detected in?
  • Are there soil-pica exposures?
  • Does the sample represent an equal size of the exposure unit?
  • Were contaminants requiring special consideration (e.g., PAHs) sampled?
When ATSDR’s Discrete and Non-Discrete Sampling Guidance Does Not Apply

ATSDR’s Discrete Sampling Guidance and Non-Discrete Sampling Guidance do not apply to all contaminants and types of sampling.

  • Consult SMEs when evaluating asbestos and lead.
  • Refer to ATSDR’s PAHs Guidance.
  • Refer to ATSDR’s Dioxins Guidance for dioxins.
  • Consult with the ADS group when evaluating dependent data: instances where the results of one sample at least partially predict the results of another sample. In these cases, data have some degree of auto-correlation, whether in space or time. For example, an exposure unit defined as a:
    • Stretch of river with known surface water contamination, surface water monitoring locations very close together would be expected to have highly correlated data that do not provide independent measures of contamination.
    • Playing field with soil contamination of a persistent pollutant, repeated sampling of the same locations over time (i.e., time series) would also be expected to have highly correlated—and not independent—measures of contamination.

General Overview of ATSDR’s Guidance for Calculating EPCs with Discrete Sampling Data

ATSDR recommends the two types of statistics listed below for estimating EPCs with discrete sampling data. The appropriate statistic depends on the site-specific exposure scenario evaluated and the characteristics of the data set.

  • Maximum detected concentration
  • 95 percent upper confidence limit of the arithmetic mean (95UCL)

When you calculate a 95UCL for discrete samples, the contaminant must meet the following criteria:

  • Analyzed in ≥ 8 samples
  • Detected in ≥ 4 samples
  • Detected in at least 20% of the samples

See the table below for more guidance on discrete sampling.

General Overview of ATSDR’s Discrete Sampling Guidance for EPC Calculations

General Overview of ATSDR’s Discrete Sampling Guidance for EPC Calculations
Exposure Duration Sample Size and Frequency of Detection Appropriate EPC
Acute Any Use the statistic (maximum detected concentration or 95 percent upper confidence limit of the arithmetic mean [95UCL]) that best aligns with the sample media and applicable toxicity data (see ATSDR’s Guidance for Discrete Sampling for further details)
Intermediate or Chronic < 8 samples
  • Use the maximum detected concentration
  • Consider recommending additional sampling
  • Consult with the Associate Director for Science (ADS) group on your evaluation
≥ 8 samples with the contaminant detected in at least four samples and detected at least 20% of the time Use the appropriate 95UCL (based on the sample size, number of detects, and data distribution)
≥ 8 samples with the contaminant detected in fewer than four samples or detected less than 20% of the time
  • Consider recommending additional sampling
  • Consult with the ADS group on your evaluation

Calculating 95UCLs

Health assessors will use a software application or programming language to calculate the 95UCL for the EPC. While many different software applications and programming languages are available, ATSDR recommends health assessors use its Exposure Point Concentration (EPC) Tool (see text box below). With the EPC Tool, you can quickly and accurately calculate EPCs following the procedures established in ATSDR’s guidance documents for calculating EPCs, which include the following:

ATSDR’s EPC Tool

ATSDR’s EPC Tool is a web application built to help ATSDR and its partner health assessors calculate EPCs for discrete environmental data during the PHA process. Health assessors can use the EPCs calculated with this tool whenever the 95UCL or maximum value are appropriate metrics for evaluating exposures. Email phast@cdc.gov to gain access to ATSDR’s EPC Tool. Health assessors can get more information on the EPC Tool from the EPC Tool User’s Guide, available on the EPC Tool Resources page.

Health assessors may also use other software applications and programming languages to calculate the 95UCL for the EPC, as listed below.

  • EPA’s ProUCL. This is a free software program. Its primary function is to calculate 95UCLs for sampling data sets. You can also access a user guide, technical guide, and training sessions.
  • R – programming language. This is a downloadable, free programming language that has many applications for conducting statistical analyses on sampling data, but requires more effort to learn compared to EPA’s ProUCL.

The initial data processing steps for 95UCL calculations are as follows:

  • Perform a data quality review.
  • Identify and process non-detect observations.
  • Identify and process lower-bound concentration data points.
  • Identify and process duplicate samples and replicate analyses.
  • Plot environmental data to visualize the measurements’ distribution.

Refer to ATSDR’s Exposure Point Concentration Guidance Documents for more information on the concepts outlined here, and on the recommended statistical tools for calculating 95UCLs and selecting the appropriate 95UCL based on the characteristics of the dataset. The ATSDR EPC Tool will calculate and select the appropriate 95UCL automatically, but health assessors should review the guidance documents to understand how the program works and address special cases that may occur for some datasets. Health assessors using either ProUCL or R should review and understand the computational procedures, steps for performing necessary quality control checks, and examples of 95UCL calculations provided in the guidance before calculating 95UCLs with either of those applications.

Handling Non-Detects

Keep in mind the following general rules for handling non-detects in your data sets:

  • Do not delete non-detect observations from data sets.
  • Do not include non-detect observations with extremely high detection limits.
  • Do not replace non-detect observations with a single surrogate value.
  • Do not calculate 95UCLs for contaminants detected in fewer than four samples or fewer than 20 percent of samples.

Calculating Soil-Pica Exposures

Health assessors should evaluate soil-pica exposures using the maximum detected concentration as the EPC. The guidance that applies to EPCs for soil-pica exposures is discussed in the Calculating Soil-Pica Exposures section.

Refer to ATSDR’s Soil and Sediment Ingestion EDG and ATSDR’s Discrete Sampling Guidance for more information.

General Overview of ATSDR’s Guidance for Calculating EPCs for Non-Discrete Sampling

ATSDR’s guidance on non-discrete sampling refers to composite sampling and the incremental sampling methodology (ISM):

  • Composite sampling involves combining multiple discrete samples for analysis. If the composite data meet the data quality objectives (DQOs) and adequately represent the exposure unit, health assessors can use the sampling results to estimate EPCs as described below.
    • Acute exposures: Examine the appropriateness of using composite sampling data to evaluate acute exposures on a case-by-case basis (see ATSDR’s Guidance for Non-Discrete Sampling). For acute examples, you must be sure that a composite sample truly represents only the area where one might have exposure over an acute duration. Refer to the guidance and contact the ADS group for help, if needed.
    • Intermediate or chronic exposures: Use a single composite sample for soil if you have determined it accurately represents the area. Refer to the guidance for more information on when using this single soil sample for intermediate or chronic exposure is appropriate. When calculating EPCs for biota sample data, a minimum of three composite samples should be collected for each species of interest. Acknowledge any uncertainty (i.e., when information is incomplete or unknown) with its use in your report. For multiple composite samples, determine whether each composite sample represents an equal size area of the exposure unit. For equal size areas, use the average as the EPC. For unequal size areas, use a weighted average as the EPC.
  • Incremental Sampling Methodology (ISM) is a structured form of composite sampling, designed to characterize the overall contamination level within an area of interest, where numerous incremental soil samples (an absolute minimum of 30; ideally at least 50 to 100) are collected and combined to form a single ISM sample for laboratory analysis. Health assessors can use the sampling results to estimate EPCs as described below.
    • Acute exposures: Use the results of a single ISM sample or average of multiple ISM samples as EPCs under certain scenarios. Refer to the guidance for more information on when using this single sample for acute exposures is appropriate.
    • Intermediate or chronic exposures: Use the result of a single ISM sample or the average of multiple ISM samples as EPCs, as long as they adequately represent the exposure unit and were collected and analyzed in a manner consistent with DQOs.

Refer to the guidance for more information on calculating EPCs for composite sampling and ISM data. Consult with the ADS group if you encounter site-specific scenarios or other conditions that are not covered in the guidance.

Handling Composite and Discrete Sampling Results

You might have a site-specific data set that contains both composite and discrete sampling results. For example, composite samples could be collected across a large residential property to evaluate average concentrations, and discrete samples could be collected from certain areas that the property owner believes are unique relative to the remainder of the property (e.g., play areas, hotspots, the center of a soccer field with barren soil).

In this case, it is not statistically appropriate to combine results collected from the two methods into a single dataset. This would weight discrete samples more heavily in the dataset, and the resulting concentration might be biased toward the average of the discrete samples. Health assessors can, however, gain valuable insights by evaluating the magnitude of difference between composite and discrete sampling data. In this situation, health assessors will need to evaluate which dataset to use on a site-by-site basis, as the most appropriate dataset will be dependent on which dataset better characterizes each exposure unit being investigated.

Page last reviewed: July 12, 2022