DRI Results and How to Interpret Them

Interpreting DRI Values

The DRI system provides three different measurements of pesticide dietary risk, as described here. The following section provides guidance on how to interpret the values generated by the DRI system via our online Lookup Tools.

How Risky is that Residue?

The DRI is a unit-less index. It is the ratio of actual pesticide exposure in one serving of food relative to the maximum, allowed daily exposures from all sources.

In the context of US law, the answer to the question “How risky is that residue?” depends on the answer to another question — “risky” relative to what?

And the “relative to what” in the U.S. is the level of dietary exposure to a pesticide that triggers EPA’s “level of concern.”

This level, in turn, is set by EPA after assessing the toxicology data on a pesticide and applying the health-driven provisions of the Food Quality Protection Act (FQPA), a law passed in 1996.

The FQPA requires EPA to take into account the special susceptibility to pesticides of pregnant women, infants, and children, and to do so by applying an additional 10-fold safety factor when setting exposure thresholds, and especially a pesticide’s chronic Reference Dose (cRfD) or chronic Population Adjusted Dose (cPAD).

EPA-set cPADs are usually one-tenth of the Agency’s corresponding cRfD. cPADs reflect the added safety factors imposed by EPA in response to the FQPA.

The added, 10-fold safety factor can be eliminated, or reduced, when the EPA Administrator determines that existing data and risk assessments adequately take into account the impact of pesticide exposures on vulnerable population groups (e.g. pregnant women, infants, and children).

For food-use pesticides, the EPA’s “level of concern” is reached when the Agency determines there is no longer a “reasonable certainty of no harm” from expected levels of pesticides in food.

There are similar chronic dietary exposure thresholds in the rest of the world, which are generally called ADIs. Chronic RfDs, PADs, and ADIs are expressed on a body-weight basis, usually in units of milligrams of pesticide per kilogram of body weight per day.

DRI-Mean and Food Supply-DRI values for a single food-pesticide combination are based on annual average residue concentrations. Some consumers will ingest food with residue levels below the “mean of the positives” level, while others will ingest higher residues. Hence, it is not possible to determine whether a food-pesticide combination with a specific DRI-M or FS-DRI value contains individual samples with residues exceeding EPA’s “level of concern.”

In addition, consumers often ingest residues of specific pesticides in multiple foods in the same day. EPA dietary risk assessments must conclude that overall pesticide exposures from all foods, and other possible routes of exposure, do not exceed the Agency’s “level of concern.”

Setting Exposure and Risk Thresholds

Establishing pesticide exposure and risk thresholds is a complex and inherently subjective task. The DRI system provides a mechanism to track dietary risk levels across large residue data sets encompassing many crops, pesticides, regions, and years.

But interpreting DRI values remains a challenge.

Food companies, consumers, public health officials, and farmers want to know how risky a given residue is. They want and need guidance in interpreting a given DRI value.

The DRI system provides reliable ways to distinguish big from modest risks, but cannot link a given residue level or DRI value to a prediction of a specific adverse human health outcome. No regulator nor pesticide risk-assessment method can do so with a high level of confidence.

Arraying DRI values along a dietary-risk continuum is one useful way to interpret DRI values.

Such a continuum encompasses very low DRIs on its left edge, to progressively higher values toward its right edge. But the challenge remains — How and where to designate zones along the continuum delineating high and possibly worrisome risks from seemingly low risks not worthy of more in-depth focus by regulators, food companies, and consumers?

Here is one way to conceptualize risk zones along the continuum:

  • A “deminimus risk” zone for very-low and zero-risk samples,
  • A “modest to moderate risk” zone with positive samples posing risks above the “deminimus risk” threshold and below the possibly significant risk, “level-of-concern” threshold, and
  • A “significant risk” zone where a combination of residue levels and frequency, coupled with pesticide toxicity, is likely to result in some consumers being exposed on some days above their personal cRfD or ADI for a given pesticide in the food they eat. Such an exposure episode exceeds the “reasonable-certainty-of-no-harm” standard governing pesticide residues in food in the US.

Remember – each individual sample of a given food is made up of about 5 pounds of product selected at random from the food supply. So, each one is actually representative of MILLIONS of servings of that food at dinner tables and school cafeterias across the country. So even a small number of high-risk samples in a commonly consumed food could pose a significant public health risk.

One Way to Set Risk Thresholds

There is no single and correct way to set “modest to moderate risk” versus “significant” or “deminimus” risk thresholds for across food-pesticide combinations.

Such choices can be informed through analyses of the distribution of residue and risk levels, coupled with policy decisions about the percent of exposure/eating episodes that can exceed a particular risk threshold.

The EPA has interpreted the FQPA’s “reasonable certainty of no harm” standard to mean that the daily exposure level for a given pesticide at the 99.9th percentile of its estimated exposure distribution curve should be below the pesticide’s cRfD or cPAD. This “Threshold of Exposure” was set forth in a March 22, 2000 Federal Register Notice, “Policy Issues Related to the Food Quality Protection Act”.

Hence, the single serving of food containing residues of a specific pesticide at the 99.9th percentile would have an individual-sample DRI ≥ 1. Such a single serving of a specific food-pesticide combination with a DRI > 1 unambiguously exceeds EPA’s “level of concern.”

It is more difficult to define risk thresholds based on DRI-M and FS-DRI levels, because both are based on mean-of-the-positives-residue levels, and provide no insights into the upper end of residue distributions. For example, a food-pesticide combination with DRI-M or FS-DRI = 0.1 could contain individual samples with DRI values over 1.

If the goal is to identify relatively high-risk food-pesticide combinations for deeper regulatory reviews and possible risk-reduction interventions, the thresholds can be aligned with regulatory goals, such as EPA’s “reasonable certainty of no harm.” The analysis can then draw on quantification of pesticide risk levels in individual samples of food in a given year or set of years, with a goal of determining what percent of samples tested — and hence what share of the overall supply of a given food — is likely to contain residues exceeding EPA’s “level of concern.”.

Uncertainty will surround any and all food-pesticide dietary risk assessment policies and procedures, whether done by EPA, the European Food Standards Agency (EFSA), or derived from DRI output tables.

But for a variety of reasons, it makes sense for regulators, the food industry, and farmers to target efforts — and investments in pest management system innovation — at those geographic regions producing food with high-risk residues in the “significant” risk zone along the DRI risk continuum.

By reducing the frequency of such episodes, progress will be made incrementally in reducing overall pesticide dietary risks.

Accounting for Residues of the Same Pesticide On or In Multiple Foods

Many pesticide active ingredients are sprayed and detected on multiple crops in a given year. But based on past work with the DRI system, it is very unlikely that a pesticide will be present in 10 different foods consumed by an individual in any given day.

So, a DRI-M limit of ~0.1 in each food should, on average, prevent excessive exposures to any given pesticide on most days.

However, for a food-pesticide combination with a DRI-M value of 0.1 or less, a few individual samples may pose risks above the maximum, acceptable DRI level of 1.0. This is because individual-sample DRI values are sometimes well over 10-fold higher than corresponding DRI-M values.

Hence, some portion of the supply of a food with a DRI-M value of 0.1 will likely have individual sample DRI values greater than 1, thereby exceeding EPA’s regulatory “level of concern.”

Pesticide laws and policy in the US and EU direct regulators to take into account cumulative exposures across multiple pesticide active ingredients in cases where structurally related active ingredients pose human risks through a common mechanism of biological action.

In response to the passage of the FQPA in 1996, the US EPA pioneered methods to conduct the first cumulative organophosphate (OP) exposure- and risk-assessment encompassing over 100 food uses of 26 OPs.

In 2000, the US-PDP reported 2,432 residues of 20 different OP insecticides in 23 US-grown foods, with an aggregate DRI-M of 30 and an aggregate FS-DRI of 2.03.

Because of these very high risks stemming from hundreds of OP residues in many common children’s foods, the EPA focused almost exclusively on the OPs in the early years of FQPA implementation.

DRI Results – Tracking Differences in Pesticide Dietary Risks

A wide array of questions can be answered through the generation of standard DRI system output tables that go much deeper than the Lookup Tools currently online.

The DRI system allows for selection of samples according to several parameters, and many combinations of parameters in both the US-PDP and UK-FSA residue data sets.

By breaking down all samples of a given food tested in a given year into specific categories of samples, output tables provide the basis for identifying differences in residue profiles and risk levels across groups of samples (e.g. sweet bell peppers grown in the U.S. versus Mexico, or conventional US peppers in contrast to organically grown US peppers).

DRI Parameters

The parameters that can be used to differentiate residues and DRI risks in one group of samples compared to other groups include:

  • Country of origin: Where the crop was grown, or for processed foods, the country where the final stage of manufacturing occurred (but not necessarily the source of individual ingredients).
  • State of origin for some samples (US-PDP only): The US state where the crop was grown or distributed (this parameter has been increasingly reported over time).
  • Geographic categories of food origin: Four reported by the US PDP— all samples regardless of country of origin; domestic samples; combined imports; and, imports from specific countries. There are four options in analyzing UK-FSA results—all samples, domestic (UK-grown), imports from other EU countries, and all other [non-EU] imports.
  • Market claim categories based on type of production: Four in the US-PDP—organic, integrated-pest-management, pesticide free, and no claim; and two in the UK-FSA—organic and no claim.

Report Options

All DRI system reports can be produced in three ways:

  1.  All samples irrespective of market claim.
  2. Organic samples only.
  3. “Conventional” samples (all samples excluding “organic”).

Rule of 10

For many imported foods in particular, relatively few samples are tested by the US-PDP or the UK-FSA in a given year, and even fewer samples are available by market claim within the samples of imported foods.

Sometimes there are too few samples tested to place confidence in the calculated mean residue level or residue frequency.

So, DRI system reports can be generated incorporating a “Rule of 10.” When invoked, the DRI system ignores all food-pesticide-country-of-origin-market-claim combinations with less than 10 samples.

An example of when the Rule of 10 should be invoked follows:

Consider a year when only three samples of organic tomatoes from Mexico were tested, and the same pesticide was found in two of the three samples.

One residue was well below the acceptable threshold in organic food (5% of the applicable tolerance), and the other was 10-times higher and clearly unacceptable based on current National Organic Program rules.

With just these two residue data points, it is impossible to determine what the true, average residue level was in all organic tomatoes from Mexico that year, and whether the low or high residue is an outlier.

Banned Organochlorines (OCs)

Residues persist in certain foods of several OC insecticides, including DDT and its metabolites, aldrin, dieldrin, heptachlor, chlordane, mirex, and toxaphene.

Most of these insecticides were banned for use on food crops during the 1970s, and few registered uses of OC insecticides remain worldwide.

As a result, there is little that farmers, regulators, the food industry, and pesticide manufacturers can do to prevent the presence and detection of banned OC residues in certain foods, especially root crops.

Two OCs remain in use in some countries and are included in the DRI system (endosulfan and methoxychlor). Because farmers have little control over OC levels in food, all DRI reports can be run with or without any reported detections of banned OCs.

Calculating DRI Values for Specific Sets of Samples

Users can access DRI tables that reflect residues and DRI values in subsets of all samples tested in a year:

  • Samples reflecting residues on imported food versus domestically grown food,
  • Food grown using a conventional, pesticide-dependent system, versus samples of food from certified organic farms,
  • Tables including all food-pesticide combinations, or just those with 10 or more samples (Rule of 10),
  • Tables including or excluding residues of banned organochlorine insecticides (e.g. DDT, aldrin, dieldrin, chlordane, toxaphene), and
  • Combinations of the above sample-selection criteria.

Levels of Aggregation

There are three forms of the DRI: DRI-Mean, the Food Supply-DRI, and individual-sample DRI values (for details of the differences among these three DRI metrics, see Three Ways to Calculate DRI Values).

Each of the ways to calculate DRI values arise from the residues reported in, and stemming from a given pesticide-food combination in a given year. Food-pesticide combination DRI values can be aggregated, or added together, in three ways in various DRI tables:

  1. All pesticides found in a given food in a given year, including details on each pesticide found in the food.
  2. Summary data reflecting all residues reported in each food tested in a given year, reporting the number of residues found in each food, the average number of residues per sample, and aggregate DRI values (but not details for each pesticide-food combination).
  3. Residues and DRI values reported in individual samples of a given food.

Tables reflecting levels of aggregation #1 and #2 above can also be generated by pesticide-food combinations. These report all the foods in which a given pesticide was found in an annual set of residue test data.

Level of Aggregation #1 is the most detailed. It is designed to easily compare the contribution of a given pesticide found in a food to total, aggregate DRI values across all the pesticides detected in the food.

These tables spotlight “risk driver” pesticides in each food — pesticides that account for the largest share of total risk from all pesticides found in the food. It is equally valuable in identifying pesticides contributing modestly, or not at all, to aggregate dietary risk.

In these 1st-level of aggregation tables, DRI values are calculated using the “mean of the positive” samples (DRI-M and FS-DRI). But these tables provide modest insight into the distribution of residue and risk levels across all the samples of a given food tested in a given year. This is a key reason why the 3rd level of aggregation is needed.

Level of Aggregation #2 provides summary metrics associated with all the pesticide residues found in each food. These 2nd-level output tables include a row of data for each of the foods tested in a given year by either the US-PDP or the UK-FSA (or set of years, e.g. all samples combined 2014-2018).

For each food-year combination, the tables report the total number of residues found, the average number of residues reported per sample, and aggregate DRI values.

The tables add up and  report total DRI values across all foods tested in a given year.  They also calculate the share of total risk accounted for by each food.

2nd-level tables drive home the fact that in most years, just a few foods account for 80% or more of total DRI risk, while one-half or more of the foods tested contribute modestly to overall risks.

Recognizing “risk driver” foods, in contrast to foods accounting for modest dietary risks, is obviously helpful in targeting pest management R+D investments and promoting adoption of prevention-based biointensive Integrated Pest Management (bioIPM) systems on conventionally managed farms.

Level of Aggregation #3 provides the most accurate estimates of pesticide risks because individual-sample DRI values are based on the actual residue levels reported in each sample, rather than the mean of the positive residues found in all samples of a given food.

The dietary risk metrics in 3rd-level tables reflect actual residue intakes by consumers from one serving of a given food. Risk levels by sample can then be ranked, highest to lowest, and the differences quantified between the riskiest samples and the least risky samples.

The bigger the difference, the greater the opportunity to lower overall risks by: (a) encouraging adoption of the IPM systems used by growers producing low-risk food, and/or (b) shifting production away from high-pest pressure, high pesticide-risk regions to production areas where bioIPM systems are proven and reliable, and farmers are far less reliant on high-risk pesticides.

3rd-level tables provide the most granular and accurate way to assess the distribution of risk levels across all samples tested of a given food. These tables also make it easy to identify “risk driver” pesticides in those samples accounting for the highest DRI values.

DRI Output Reports

The DRI system is programmed to generate thousands of standard reports. Each DRI system report is composed of several annual versions of the same DRI output table.

Annual output tables are organized in a DRI report from the most-recent year for which data are accessible, to the first year results were reported (usually 1993 through 2018 in the case of the PDP, and 1991 through 2019 for the UK-FSA data).

Summary statistics in the bottom rows of all output tables show, among other things, the total number of residues found, the average number of residues per sample tested, and aggregate totals for DRI-M and FS-DRI.

Many DRI system output tables and reports show the share of aggregate FS-DRI that is accounted for by each food-pesticide combination. The foods and pesticides in DRI output tables are typically ranked from the largest to smallest aggregate FS-DRI value, and hence highest to lowest share of overall risk.

By doing so, it is easy to see the risk-driver pesticides in every food, and the degree to which they account for aggregate DRI risk totals. It is also often clear whether or not the risk-driver pesticides in a given food are worth worrying about.

For example, in many foods tested by both the US-PDP and UK-FSA in any given year, the top risk-driver pesticide has a FS-DRI value of 0.01 or less — not very worrisome.

While uncertainty remains in just how risky those residues are, investing time and resources in reducing exposures to these food-pesticide combinations is far less important than targeting other foods, for which risk-driver, food-pesticide combinations have FS-DRI value of 0.5, or higher.

Displaying DRI output tables in this way also provides insight into risk distributions. Usually 10% or fewer of the residues reported in a given food account for 95% or more of aggregate FS-DRI.

Likewise, when results are reported by pesticide, just a few foods account for the majority of aggregate, pesticide specific DRI risk levels from reported residues of a specific pesticide across foods.

Drawing on either US-PDP or UK-FSA residue data, the DRI system produces multiple versions of 6 standard output tables and reports (numbered 3–8 in the list below).

Each report can be generated in Microsoft Excel, Access, or online. Thousands of annual output tables are currently available for 1993 through 2018 based on US-PDP data, and for 1999 through 2019 based on UK-FSA data.

The standard, 8 DRI output reports include:

Report 1 – Chemical Names, Classifications and EPA or EFSA Toxicity Benchmarks (cRfDs, ADIs).
Report 2 – Number of Samples by Food, Country of Origin, Market Claim, and Year.
Report 3 – Food-pesticide Combination DRI Values by Year, Ranked by FS-DRI.
Report 4 – Pesticide-food Combination DRI Values by Year, Ranked by FS-DRI.
Report 5 – Aggregate Food DRI Values by Year, Ranked by FS-DRI across All Pesticides.
Report 6 – Aggregate Pesticide DRI Values by Year, Ranked by FS-DRI across All Foods.
Report 7 – Number of Residues and Aggregate DRI Values by Type of Pesticide and All Pesticides by Year.
Report 8 – Number of Residues and Aggregate DRI Values by Food, Food Form, and Food Group by Year (under development).

The DRI system generates multiple versions of standard output tables and reports according to combinations of the country-of-origin of the samples tested, market claim, and inclusion criteria (Rule of 10 and OCs in or out).

There are seven country-of-origin options for each report:

  1. All positive samples, regardless of country of origin;
  2. All imported samples;
  3. Imported samples disaggregated by country of origin;
  4. US Domestic samples;
  5. UK domestic samples (grown or processed);
  6. Imported samples into the UK from other EC countries; and
  7. Imported samples into the UK from non-EC countries.

The multiple versions of standard reports 2–8 also can include 3 market-claim options:

  1. All samples, regardless of market claim;
  2. Samples lacking a market claim, and referred to as “Conventionally Grown”; and
  3. Samples labeled as organically grown.

All standard DRI output tables and reports can be run with or without the Rule of 10, and with or without residues of banned OC insecticides.

Accordingly, the DRI system currently generates 960 reports, each with multiple output tables.

Of this total, there are 384 US-PDP reports [8 standard reports × 4 country of origin options × 3 market claim options × 4 inclusion criteria options (Rule of 10 option in or out, banned OCs in or out)].

The system currently produces 576 reports drawing on UK-FSA data (8 × 6 × 3 × 4).

Each of these 960 US-PDP and UK-FSA based output reports contains, on average, annual output tables covering about 25 foods.

For each food-pesticide combination, the residue testing programs average around 5 annual output tables in an output report, reflecting the 5 years over the span of program years that the food was contained in the testing program, or on average about 125 food-year output tables.

In the case of pesticide-food combinations, there are about 40 pesticide-food output tables each year, or an average total of around 1,000 over the span of US-PDP and UK-FSA testing. Each such table presents the foods tested in a given year that contained residues of the pesticide.

On average across the DRI system, there are about 500 annual output tables produced per DRI system report.

Accordingly, the DRI system generates roughly 480,000 annual output tables (960 output reports containing, on average, 500 annual output tables).

In addition, the DRI system produces thousands of tables reporting individual-sample results.

As new analytical applications are carried out, the number and diversity of standard DRI output tables and reports will increase.

Limits of the DRI System

DRI values are only as accurate as the data used to compute them.

In general, greater confidence can be placed in the DRI’s residue-data-driven numerator, than in the DRI denominator, which is calculated from a pesticide active ingredient’s cRfD/cPAD/cADI.

High quality, extensive pesticide residue data are now available from the US-PDP and UK-FSA and have been incorporated in the DRI system. Both testing programs emphasize frequently consumed foods that typically contain residues.

Both data sets permit risk analyses at several levels of aggregation, as well as risk levels in samples from specific geographic regions and in food harvested from organic vs. conventional production systems.

Individual sample-specific results show the distribution of residue levels and DRI values across all samples tested in a given year.

The risk-assessment methods used by EPA and European regulators produce a single estimate of a pesticide’s chronic toxicity, based on the adverse response that occurs at the lowest dose among all toxic effects observed.

Some pesticides pose one, or just a few types of risk, the worst of which may quickly dissipate soon after exposures end.

Other pesticides may cause several possible adverse effects, some of which may be essentially irreversible. Some types of adverse effects are well explored, while other, emerging toxicity endpoints are not (e.g., impacts on the microbiome and epigenetic effects).

Most EPA-set cRfDs and EFSA’s chronic or acute ADIs do not take account of the diversity of adverse biological outcomes that a given pesticide might cause.

However, a current cRfD or cADI might be protective of other adverse biological responses following exposure to a pesticide, if the current cRfD or cADI is actually based on the most sensitive endpoint.

Regulators have the tools needed to insure that existing risk thresholds are rarely, if ever, exceeded, if they choose to use them. The fact that other risks may exist from exposures to certain pesticides does not diminish the need to mitigate known risks.

See more: