Tell the FDA: Toxic chemicals shouldn’t be in our food.

Unleaded Juice: Getting Credible Lab Results is Essential

EDF (August 2, 2022)

Originally published here.

FDA’s move to establish action levels on lead in juice – and eventually other foods that young children eat or drink – is an important step forward. While we believe that the action levels need to be tougher, any action level has a limited value if labs that analyze samples for contamination provide results that buyers, regulators, or consumers cannot trust.

We recommend that labs meet four criteria to provide credible results:

  • Be accredited under international standards for testing and calibration of labs (ISO/IEC 17025);
  • Use the analytical method based on FDA’s Method 4.7 [PDF, 1.16MB];
  • Demonstrate proficiency in a third-party, blinded test to quantify lead, arsenic, and cadmium to around 6 parts per billion (ppb); and
  • Provide a written report of results at that level.

Here is the list of labs that met these criteria as of August 2021. See below for our in-depth analysis.

In our experience, the third criterion is the most challenging. As described in detail below, Fapas, a company that performs third-party proficiency testing internationally, conducted a blinded test with baby food samples containing lead, arsenic, and cadmium at around 6 ppb. Almost half of the labs failed to accurately measure the toxic elements at that level.

A second round of testing yielded similar results for labs that either failed the first round or participated for the first time. Fapas is conducting another round of proficiency testing starting August 8, 2022. We encourage stakeholders to ensure participation of any labs on which you might rely.

We also found that labs demonstrating proficiency at around 6 ppb needed to upgrade their lab reports to meet the fourth criterion. Some reported levels as high as 10, 15, or 20 ppb as nondetectable, despite being able to do better. Only when challenged to lower the reporting limit did they upgrade their reports and update their accreditation. If you see a reporting level over 6 ppb, consider a different lab.

To find a list of labs that met the four criteria, we encourage you to use the list published by the Baby Food Council as of August 2021. We helped develop the criteria and evaluate the labs before EDF withdrew from the Council in October 2021. We are working with Fapas and other stakeholders to update the list.

A lab needs to measure contaminants at levels much lower than FDA’s action level.

Well-managed companies strive to stay well below the action level to avoid risking a recall. Beyond the costs of recalling products, the reputational risk can have long-term impacts.

Therefore, we recommend labs that can reliably measure contamination at levels at least 3 – and ideally 10 – times lower than the maximum allowed concentration (i.e., the action level). With this type of margin, companies can identify and fix potential problems with a supplier before having to consider a recall.

This approach is consistent with FDA’s Preventive Control Rule. It requires food facilities to conduct a hazard analysis and implement risk-based preventive controls to avoid hazards such as pathogens or heavy metals in their products. Food facilities must have a written Food Safety Plan [PDF, 86.5KB] so they – and regulators – know how they will comply with the rule.

Is a reporting limit of 6 ppb sufficient?

FDA has proposed a lead action level of 10 ppb in apple juice and 20 ppb in other juices. Almost a decade ago, it proposed an action level of 10 ppb for inorganic arsenic in apple juice. We don’t know what FDA will propose for other children’s food, but Europe has set a limit of 20 ppb for baby foods. Based on these limits, a lab with a 6-ppb reporting limit for lead, arsenic, and other toxic elements may be sufficient for most foods, but it is cutting it close for apple juice.

When Fapas conducted its Round 1 blinded proficiency testing [PDF, 1.06MB] for vegetable purees[1] in 2020, it prepared samples in three tiers aiming for roughly 1.5 ppb, 7 ppb, and 12 ppb. It sent the blinded samples to 28 participating labs that analyzed the samples and reported the results to Fapas, who compared the results statistically. Note that the actual levels varied for each toxic element and are being reported only as an approximation here.

2020 Round 1 Blinded Proficiency Testing Results for 28 Labs
Percentage of Labs Detecting at specified ppb
Contaminants~1.5 ppb~7ppb~12ppb
Lead alone37%70%81%
Lead, Cadmium & Arsenic26%52%67%

With so many of the 28 participating labs failing at the lowest concentration, 1.5 ppb was not a practical option. After consulting with the labs that demonstrated proficiency at 7 ppb, the Baby Food Council settled on 6 ppb as a realistic measure.

Note that labs should be challenged to do better than 6 ppb. When Healthy Babies Bright Futures or EDF wants to analyze samples for lead, arsenic, or cadmium, we send samples to Brooks Applied Labs. For a fruit puree, the lab has a limit of quantification (LOQ) of 3 ppb and a limit of detection (LOD) of 0.9 ppb.

A closer look at the first two criteria for credible labs.

Our four criteria are essentially the same as those the Baby Food Council developed when EDF was a member. In the discussion above, we have focused on the third and fourth criteria, but the first two are also important.

The first criterion is that labs be accredited under the latest version of ISO/IEC 17025:2017 for the testing and calibration of laboratories. The standard is designed to enable labs “to demonstrate that they operate competently and generate valid results.” ISO is the International Organization for Standards, an independent, nongovernmental, international organization for national standards bodies like the American National Standards Institute (ANSI) in the United States.

Almost all commercial labs have this accreditation. However, we have found that in-house labs are often not accredited. From our perspective, the ISO standard is essential to ensure that a lab demonstrating proficiency is also able to consistently report accurate results.

The second criterion is that labs must use the most sensitive method to analyze samples. Currently, this technique is FDA’s Method 4.7 as part of the Elemental Analysis Manual for Food and Related Products. FDA reports that method has been validated at 14 labs (9 state labs and 5 federal labs) and has a LOQ of 10.9 ppb and an LOD of 1.20 ppb for lead in 25 foods. We found that some commercial labs had refined the method further and can provide accurate measurements at lower levels for the simpler foods that babies commonly eat.

If the results are being used for regulatory compliance, we encourage people to ask the lab for detailed  analysis of its accuracy and precision for the samples. Regulators may want this information if questions arise.

FDA’s Lab Oversight Program

Recognizing the challenge of food companies ensuring a lab provides credible results, FDA launched a Laboratory Accreditation for Analysis of Foods (LAAF) program this year. Food testing labs and the accreditation bodies that evaluate them under programs like ISO/IEC 17025 can voluntarily participate in the program. In July 2022, FDA registered six accreditation bodies under the program – an important step forward.

If food testing is required under the rules, then companies must use a participating lab. While food testing is generally not required for action levels like those for lead in juice, we encourage companies to use a participating lab, especially as the LAAF program develops.

Summary

Food companies, regulators, and consumer advocates should be relying on labs that meet the four criteria described above to avoid risk of product recalls, potential reputational damage, and loss of consumer trust. At the current time, they can refer to the Baby Food Council’s list of labs and look for updates from Fapas. In addition, they should be challenging labs to report at lower concentrations of lead, arsenic, and cadmium and—as the opportunity arises—to participate in FDA’s LAAF program.

[1] Vegetable puree would be comparable to fruit juice.