Deep Dives: Environmental Health Policy

Tom Neltner, Senior Director, Safer Chemicals Initiative

Note: This is the third of three blogs in this series. Click here for Blog 1 or Blog 2.

On January 24, 2023, FDA released draft action levels for lead in foods intended for babies and young toddlers.¹ For most of these foods, the limit would be 10 parts per billion (ppb). FDA proposed a less-protective limit of 20 ppb for dry infant cereals and single-ingredient root vegetable products (i.e., carrots or sweet potatoes) because these products may have greater lead contamination levels. This proposal is a key step in implementing the agency’s Closer to Zero Action Plan.

In this blog, we take a deeper dive into the research needed for FDA to fulfill its commitment to continuously improve its action levels. We identify four issues for which FDA, USDA, and industry should support research if we are to make progress in reducing toxic element contamination of food:

Rice is highly contaminated with lead, as well as with high levels of inorganic arsenic. Rice has long-been recognized as the primary source of young children’s dietary exposure to inorganic arsenic. FDA’s data² now make clear that lead contamination is also a significant issue. For the 299 rice cereal samples, the 90thpercentile was 21 ppb of lead, and 28% had levels of 10 ppb or more. In contrast, only 1 of 85 samples (1.2%) for non-rice or multigrain cereals had 10 ppb or more of lead, and the 90th percentile was 8 ppb, so a 10-ppb level can be met and is justified.

In 2020, FDA highlighted that inorganic arsenic concentrations for infant rice cereal had dropped from 64% greater than 100 ppb in 2011-13 to 53% in 2014 to 26% in 2018. Some of this progress can be attributed to baby food companies screening out rice with higher levels, a change that shifts the more highly contaminated products to general market. However, there is also compelling research showing that improved growing and processing methods such as intermittent flooding of fields, removing bran, and decanting cooking water significantly reduce inorganic arsenic contamination.

More research is needed to reduce lead contamination in rice and how it correlates to inorganic arsenic levels). We also need additional research on essential nutrients in rice.

Sweet potatoes commonly have significant lead contamination, but research shows promise. Lead contamination of sweet potatoes was significant enough that FDA set an action level for foods where they were the only ingredient. FDA’s three data sources evaluated 14 samples of sweet potato puree and half of the samples were above 15 ppb with the highest at 34 ppb.

Several years ago, EDF funded Dr. Arthur Villordon at Louisiana State University’s Sweet Potato Research Center to investigate the lead contamination and identify potential research opportunities. He and his team determined that much of the contamination was in the peel, suggesting that optimized peeling would be beneficial. They also identified the possible role of type of sweet potato—as well as opportunities to manage essential nutrient availability—to reduce uptake of lead.

Along with researchers from Mississippi State University, University of California—Davis, Colorado State University, and two units of USDA’s Agricultural Research Service, Dr. Villordon and his team applied to USDA last month for significant research support to evaluate those and other opportunities.

Quinoa is highly contaminated with lead and cadmium. Arsenic levels were not provided. The levels of lead in quinoa were significantly higher than rice cereal, grain snacks, or root vegetables. While the data set is older (from 2014) and much smaller (29 samples), the 90th percentile was an astounding 90 ppb of lead. In addition, 41% had 20 ppb or more of lead and 58% were over 10 ppb.³ In addition, cadmium was also present in all but two of the samples ≥ 10 ppb and 75% were ≥ 50 ppb.

EDF talked with one of the leading researchers on quinoa, Dr. Lori Hoagland of Purdue University, about opportunities to reduce contamination. She thought that changing strains of quinoa might be the best strategy, although research funds are limited.

Carrots also have significant lead contamination. As with sweet potatoes, FDA data showed that carrots were also contaminated with lead, although at lower levels than sweet potatoes—the other root vegetable common in baby food. For the 39 samples of carrot puree, the 90th percentile was 15 ppb of lead, and 20% had levels of 10 ppb or more.

Several years ago, EDF funded Dr. Hoagland to investigate lead contamination in carrots and identify potential research opportunities. She and her team determined that much of the contamination was in the outer areas of the carrots, suggesting that optimized peeling would be beneficial. They also found that a particular strain of carrot showed promise in reducing uptake of lead and cadmium.

Along with researchers from Michigan State University, Cornell University, University of Buffalo, University of California—Davis, and a unit of USDA’s Agricultural Research Service, Dr. Hoagland and her team applied to USDA last month for significant research support to determine more effective ways of reducing lead in carrots.

Additive or Synergistic Impacts of Lead, Cadmium, Arsenic, and Mercury
FDA’s Closer to Zero Action Plan focuses on four toxic elements—lead, cadmium, inorganic arsenic, and mercury. One of the early goals was to consider the cumulative effects of these toxic elements on children’s developing brains.

Dr. Piper Reid Hunt and a team at FDA’s Office of Applied Research and Safety Assessment have been studying the harm to the developing nervous systems of nematodes (C elegans) because they are similar enough to human neural development to allow helpful insights. Her important research is not yet made public.

In summary, we encourage FDA, USDA, and industry to invest in research to reduce contamination of foods with toxic elements and evaluate the research that the presence of these substances in the diet can have on children.

Sign up to receive notifications when we publish a new Deep Dives post!

¹ We included all 29 samples that FDA identify as quinoa, even if the agency did not also identify it as a dry cereal.

² 1) Toxic Element Program consisting of 356 samples of dry infant cereals, fruits, mixtures and vegetables collected from 2008-2021; 2) FDA Survey 1 consisting of 147 samples of dry infant cereals and mixtures collected from 2013 to 2014; and FDA Survey 2 consisting of 360 samples of fruits, mixtures, vegetables, yogurts, custards/puddings, and single-ingredient meats collected in 2021.

³ FDA’s guidance refers to babies and young children. However, the action levels only apply to foods intended for children younger than 24 months, a very narrow definition of young children, especially since children up to six years of age are particularly vulnerable to the harm to their brain from lead. Other federal agencies set standards for this broader age range. Toddlers are generally considered to be between 1 and 3 years of age. Therefore, we use the term “toddlers” to avoid confusion.

Tom Neltner, Senior Director, Safer Chemicals Initiative

Note: This is the second of three blogs in this series. Click here for Blog 1 or Blog 3.

On January 24, FDA released draft action levels for lead in foods intended for babies and young toddlers.¹ For most of these foods, the  limit would be 10 parts per billion (ppb). FDA proposed a less-protective limit of 20 ppb for dry infant cereals and single-ingredient root vegetable products (i.e., carrots or sweet potatoes) because these products may have greater lead contamination levels. This proposal is a key step in implementing the agency’s Closer to Zero Action Plan.

Overall, FDA fixed several problems we highlighted in our Unleaded Juice series about the agency’s draft action levels for lead in juice, which it released in April 2022. However, significant issues remain, especially regarding the agency’s transparency, that undermine the credibility—and ultimately the effectiveness—of the action levels.

We discuss the proposed action levels and recommendations for more protective limits as well as specific research needs in separate blogs. Here, we take a deeper dive into FDA’s approach to setting these action levels.

Action Levels v. Regulatory Standards
FDA’s regulation at 21 C.F.R. § 109.6 addresses unavoidable contaminants in foods. It says FDA should use action levels when “technological or other changes that might affect the appropriateness of the tolerance are foreseeable in the near future.” If the agency does not foresee changes that would reduce the contamination, the regulation indicates it should set tolerances through rulemaking. By issuing action levels, FDA is affirming its commitment to continuous improvement —a key element in its Closer to Zero Action Plan.

In general, FDA develops action levels through guidance rather than rulemaking because it claims that the process enables the agency to move more quickly and to make updates more easily. Guidance has typically been subject   to less stringent interagency review by the White House’s Office of Management and Budget (OMB).²

Nonetheless, FDA’s action levels for lead and inorganic arsenic in food have undergone OMB review, and that review has been quite lengthy.³ The delays have been significant enough that FDA revised its Closer to Zero Action Plan in January by:

• Changing its commitment to publish draft action levels for lead and arsenic to only submitting the document for interagency review. These documents are not public until finalized.
• Eliminating its commitments to finalize all of its draft action levels.
• Dropping the word “Action” from the title of the program.

Lack of Transparency in Decision-Making Undermines Credibility and Effectiveness
We do not suggest that action levels need to undergo the full regulatory interagency review by OMB. However, there are elements of that review that would improve the transparency and credibility of the decision, as well as the effectiveness of the action levels. They might also accelerate OMB review. Therefore, we recommend:

Explaining the options the agency considered and why it selected the action level described in the guidance.
FDA’s current approach only compares the selected option to the status quo. We know many of the excellent FDA staff working on the issue and how seriously they approach this effort. They surely considered many variations before settling on the proposed option. However, that information is largely omitted from the proposed guidance and supporting materials. This omission makes it more difficult to provide constructive comments to the agency and undermines the credibility of the decision.

Estimating dietary exposures that protect more than just the 90^th percentile of children. FDA typically uses the 90^th percentile to evaluate exposure to contaminants or additives. We have not seen a rationale for selecting this value over more protective ones—like the 95^th percentile commonly used in Europe for food or, in the U.S., using the 97.5^th percentile to set an elevated blood lead level (CDC) or the 99^th percentile for pesticides in the U.S.

As we explained in a blog on the lead-in-juice proposal, while percentiles may seem abstract, we think it is important to recognize that, in this case, they represent lead-exposed children. Using the 90th percentile means that the 2.4 million children ages 1 to 6 years who are exposed to higher levels of lead are not taken into account.⁴ When including children younger than age 1, the number is even greater.

Calculating the socioeconomic benefits of the options considered and the one selected. As we explained in a blog on the lead-in-juice proposal, we encourage FDA to use methods developed by EPA (and accepted by OMB) to quantify the societal benefits of reducing young children’s exposure to lead. We applied that method to estimate that a 6% reduction in exposure for children younger than six years of age would yield $1 billion per year in benefits.We think this approach is a valuable tool to help the agency and stakeholders consider options. It would also prompt industry to provide realistic estimates of the cost to achieve the limits, rather than the vague claims it often relies on.

Transparently comparing the options considered to their impact on FDA’s Interim Reference Level (IRL) of 2.2 µg/day for children. In 2018, FDA established an IRL of 3.5 µg/day. The agency describes this value as the “maximum daily intake for lead from food” of lead for children. This is not a health-based level—rather it is a target the agency uses to drive lead contamination lower through its 2021 Closer to Zero Action Plan. Using FDA’s own analysis, we estimated that more than 2 million young children (aged 1-6 years) exceeded the IRL.

In July 2022, FDA lowered the IRL to 2.2 µg/day to conform to changes in CDC’s Blood Lead Reference Level. We estimated that 7 million young children (one-third of all children in this age group) exceeded this level and called on FDA to more rigorously compare its proposal—including options considered—to the IRL. We also asked the agency to estimate the number of children who would no longer exceed the IRL as a result of each option.In the draft guidance document, FDA states that under the status quo “the 90th percentile dietary exposures for babies and young children are below the IRL for lead of 2.2 μg/day for children.” While the agency shows that 90th percentile intake of specific groupings of foods are below the IRL, the agency provides no details for how it reached that conclusion, which should apply to dietary intake from all sources and not just those covered by the action levels.⁵

In other words, the approach would exclude the contribution from food not covered by the action levels, such as: 1) juices; 2) grain snacks (e.g., teething biscuits, cookies, and puffs); 3) fruits, cereals, and vegetables that parents use to make homemade foods for their babies and young toddlers, even though these products are often likely to be more contaminated, regardless of whether they are organic; and 4) foods marketed for a general audience such as applesauce and canned food, that parents (especially those on tight budgets) commonly feed to children.

In addition, the Agency’s assertion contradicts that of an FDA scientist who estimated in 2019 that children 12 to 36 months had a 90th percentile lead intake of 2.6 μg/day from their diet.⁶ It is unclear if this is due to differences in methodology, or a true decrease in exposure between then and now.

Assuming the claim is true and that the 90th percentile is at 2.2 μg/day, it still means that 10% of the approximately four million children born each year— or about 400,000 children – exceed the IRL. This illustrates the flaw in choosing the 90th percentile to represent an “upper bound” and “a health protective measure to account for babies and young children (0-23 months) who consume larger amounts of food and would therefore have higher exposures.”

As we have previously commented, FDA’s choice of the 90th percentile is out of step with other agencies, including CDC and EPA. The agency should not only be transparent about its calculations, but it should be clear how many children will be brought below the IRL with each option considered in order to provide critical context for the strengths and weaknesses of the agency’s analysis.

Posting all toxic element testing results—not just lead—for the data supporting the guidance. When FDA evaluates a sample for lead, the lab method also provides results for cadmium, arsenic, and mercury. This is important since FDA’s Closer to Zero Action Plan is focused on the health risks to young children from lead, cadmium, mercury, and inorganic arsenic and highlights the cumulative risks from these toxic elements.

Unfortunately, FDA only provides the results for lead when it posts the supporting data for both its draft action levels for lead in juice and lead in food intended for babies and young toddlers. The agency provides no explanation for why it withholds data for the other contaminants.We raised this concern in our comments on the draft action levels for lead in juice. FDA failed to address its lack of transparency despite the opportunity and minimal burden.

Posting a red-lined version of the document that compares the original version submitted to OMB for review and the final version. Agencies, including EPA and FDA, post a comparison that shows the changes made to its rule and accompanying justification as a result of the OMB review. FDA does not do that for OMB review of guidance. The comparison helps stakeholders understand the key issues, and it poses little burden to either agency. FDA should post the red-lined version.

In summary, FDA needs improve the process by which it proposes and finalizes action levels by improving its transparency. This will make the agency’s decisions more credible—an important consideration for ensuring industry compliance with the guidance.

Up Next: Unleading Baby Food: FDA, USDA, and industry need to invest in research to reduce lead in baby food

Sign up to receive notifications when we publish a new Deep Dives post!

NOTES

1 FDA’s guidance refers to babies and young children. However, the action levels only apply to foods intended for children younger than 24 months, a very narrow definition of young children, especially since children up to six years of age are particularly vulnerable to the harm to their brain from lead. Other federal agencies set standards for this broader age range. Toddlers are generally considered to be between 1 and 3 years of age. Therefore, we use the term “toddlers” to avoid confusion.

2 Specifically, OMB’s Office of Information and Regulatory Affairs (OIRA).

3 For example: 1) Lead in food intended for babies and young toddlers where proposed guidance was under OMB review for 9 months; 2) Lead in juice where proposed guidance was under OMB review for 13 months; 3) Inorganic arsenic in infant rice cereal where final guidance was under OMB review for 4 months; and 4) Inorganic arsenic in apple juice where final guidance has been under OMB review for 21 months and counting.

4 Approximately 4 million children are born each year in the United States. Over a six-year period, that would mean ~24 million children in the age range.

5 The three values in the sixth column of Table 4 add up to 2.02 μg/day but that is not a valid method for 90th percentiles.

6 Based on the hybrid estimate from the report. Note that the lower bound is 1.8 and upper bound 4.4 μg/day.

Tom Neltner, Senior Director, Safer Chemicals Initiative

Note: This is the first of three blogs in this series. Click here for Blog 2 or Blog 3.

On January 24, 2023, FDA released draft action levels¹ for lead in foods intended for babies and young toddlers.² This proposal is a key step in implementing the agency’s Closer to Zero Action Plan.

For most of these foods, the lead limit would be 10 parts per billion (ppb). FDA proposed a 20-ppb limit for dry infant cereals and single-ingredient root vegetable products (i.e., carrots and sweet potatoes), because these products may have greater lead contamination levels.

The action levels do not apply to five types of foods that young children frequently eat:

• Grain-based snacks such as teething biscuits, sandwich cookies, and puffs—despite FDA’s finding that the biscuits and cookies were two of the three foods with the highest mean lead concentrations that the agency tested for its Total Diet Study from 2018-2020.
• Fruits, cereals, and vegetables that parents use to make homemade foods for their babies and young toddlers—despite those products being likely to be more contaminated, whether or not it was labelled as organic.
• Food marketed for a general audience (e.g., raisins, applesauce, canned food) that parents commonly feed to children —especially families on tight budgets.
• Food marketed to young children over age 2, including kids’ meals and some snack bars.
• Juices, since FDA proposed limits on those in April 2022 and should finalize them this year.

FDA has not set—or even proposed—action levels for lead for any of these foods (except juice). Based on the current Closer to Zero Action Plan schedule, the agency has no plans to establish action levels for the foods listed in items 2 and 3, above.

Do Food Companies Have to Comply with Action Levels?
The short answer is technically and legally “No”—but in practice, “Yes”…with important caveats. As with all guidance, FDA makes clear that:

[G]uidance documents do not establish legally enforceable responsibilities. Instead, guidances describe FDA’s current thinking on a topic and should be viewed only as recommendations, unless specific regulatory or statutory requirements are cited. The use of the word should in FDA guidances means that something is suggested or recommended, but not required.

As a practical matter, action levels serve as limits that companies meet to decrease the risk of having FDA deem their products adulterated and subject to a recall. In addition, FDA expects food facilities to use their Hazard Analysis and Preventive Control Program to achieve action levels by managing their facilities and suppliers to ensure compliance and taking action if problems are revealed through testing or other efforts.

FDA’s Criteria for Setting Action Levels
There is no safe level of lead in the diet. Lead can enter the food chain through multiple sources, including crops grown in contaminated soil and/or irrigated with contaminated water, atmospheric deposition from industrial activities, and old food-processing equipment that contains lead. The objective is to get exposure closer to zero.

In its proposed guidance, FDA said:

[W]hen evaluating possible action levels under 21 CFR 109.6 for lead in foods intended for babies and young children less than two years old, we took into account several considerations, including:

• the action level should minimize the likelihood that a consumer will be exposed to lead levels exceeding the IRL (interim reference level);
• as appropriate, there should be a limited number of unique action levels for simplicity;
• the action levels should result in a reduction in exposure to lead; and
• for those baby foods where lead levels are already relatively low, the action levels should be established where achievability is in the 90th-95th percentile range.

EDF’s Recommendations for Making FDA’s Proposed Action Levels More Protective
We did a deeper dive into FDA’s proposal and the three datasets³ it published to support its decision. We applied FDA’s achievability target of 90th percentile to develop four recommendations to improve the action-level proposal and reduce dietary exposure to lead for babies and young toddlers, even though we think FDA should use a stricter target for infants and babies—as it did for inorganic arsenic in infant rice cereal.⁴ For the sake of simplicity, we also sought to minimize the number of unique action levels:

1. The 20-ppb action level for dry infant cereals should be limited to rice cereals; other grains should have a 10-ppb action level. For the 299 rice cereal samples, the 90th percentile was 21 ppb of lead, and 28% had levels of 10 ppb or more. By contrast, only 1 of 85 samples (1.2%) for non-rice or multigrain cereals had 10 ppb or more of lead, and the 90th percentile was 8 ppb—so a 10-ppb level can be met and is justified. Note that rice cereal is also high in inorganic arsenic.
2. Beyond dry infant cereals, foods that contain no root vegetables as ingredients should have an action level of 5 ppb instead of 10 ppb. For 478 samples without root vegetables⁵, the 90th percentile was 3.3 ppb of lead, while 6% had 5 ppb or more. Note that half of those over 5 ppb had quinoa as an ingredient. Based on these results, foods other than dry cereals that do not have root vegetables should have an action level of 5 ppb. For 110 samples of food with root vegetables as one of multiple ingredients, the 90th percentile was 11 ppb of lead and 22% had 5 ppb or more. For foods whose only ingredient is a root vegetable, a 20-ppb action level is justified.
3. Grain snacks, such as teething biscuits, puffs, snack bars, and cookies should have an action level of 20 ppb; they are some of the most contaminated baby foods. For 123 samples of these foods collected in 2013-14, the 90th percentile was 18 ppb of lead and 26% had 5 ppb or more. While the samples are old, there are sufficient samples of these popular products with high enough levels that the agency should set a level of 20 ppb.
4. Quinoa, an increasingly popular ingredient for baby food, requires greater scrutiny, starting with increased sampling, since almost half of the samples with quinoa as the sole or main ingredient have lead levels of 20 ppb or more. The levels of lead in quinoa were worse than rice cereal, grain snacks, or root vegetables. While the data set is older (from 2014) and much smaller (29 samples), the 90th percentile was an astounding 90 ppb of lead. In addition, 41% had 20 ppb or more of lead and 58% were over 10 ppb.⁶ Without more data, FDA should maintain a limit of 20 ppb for dry cereal and 10 ppb if that cereal is combined with other ingredients.

We made these recommendations in our formal comments on the proposal that we submitted on March 27, 2023 . In subsequent blogs, we identify significant problems with the process FDA uses to establish action levels for lead and to describe specific research needs. (Updated on 3/28/23 to reflect submission of our comments to FDA and link to the document.)

Up Next: Unleading Baby Food: FDA needs to improve how it sets action levels to be more transparent and credible

Sign up to receive notifications when we publish a new Deep Dives post!

NOTES

¹ FDA’s guidance refers to babies and young children. However, the action levels only apply to foods intended for children younger than 24 months, a very narrow definition of young children, especially since children up to six years of age are particularly vulnerable to the harm to their brain from lead. Other federal agencies set standards for this broader age range. Toddlers are generally considered to be between 1 and 3 years of age. Therefore, we use the term “toddlers” to avoid confusion.

² Action levels represent levels at which FDA may regard food as adulterated and seek a recall.

³ In its 2018 sample results, 29% of the 149 samples exceeded 100 ppb of inorganic arsenic in infant rice cereal.

⁴ 1) Toxic Element Program consisting of 356 samples of dry infant cereals, fruits, mixtures and vegetables collected from 2008-2021; 2) FDA Survey 1 consisting of 147 samples of dry infant cereals and mixtures collected from 2013 to 2014; and FDA Survey 2 consisting of 360 samples of fruits, mixtures, vegetables, yogurts, custards/puddings, and single-ingredient meats collected in 2021.

⁵ Excluding dry infant cereals.

⁶ We included all 29 samples that FDA identified as quinoa, even if the agency did not also identify them as  dry cereal.