Gluten-Free Cheerios: Take Two
Update August 14, 2015
General Mills and I spoke via phone on August 13th. Here are some of the positives…
- They are looking to source some oats from suppliers who follow a purity protocol.
Note: I offered to help facilitate this process by reaching out to my contacts that supply pure oats to manufacturers. General Mills accepted this offer. Three suppliers were contacted this morning.
Update: As of 8/27/15 email correspondence with General Mills on whether they were truly interested in purchasing pure oats for Cheerios, they responded that, “It is not accurate to say that General Mills is not interested in purchasing pure oats for gluten free Cheerios and other products” and “at this time you should not have any expectation that Cheerios will be moving to a purity protocol.”
- They will be sourcing a large percentage of oats from farmers who do not grow barley or have not grown barley in the past few years.
- They will be testing more individual boxes of Cheerios.
All of the above is great news and will lead to a safer product in my opinion. The less barley in the oats to begin with, the less barley there will be in the final product.
Here are some of the not-so-positives (in my opinion)…
- They will continue to use lot means based on testing a composite sample comprised of 12 to 18 boxes to help determine the safety of final product Cheerios.
General Mills pulls 12 to 18 boxes during a 24 hour production cycle, grinds the contents of each box, takes a sample from the contents of each box, mixes all the samples together and regrinds, tests 12 sub-samples, and averages the results to provide a lot mean.
Why this bothers me: When I visited General Mills I was shown extraction values for some of the lots of yellow box Cheerios produced during the gluten-free validation phase (these Cheerios did NOT go into boxes labeled gluten-free). One extraction from one of the lots was above 90 parts per million of gluten. Because the other extraction values were so low, the lot mean was somewhere in the range of 10 to 13 ppm gluten. In my opinion if test results on a composite sample include values at or above 20 ppm gluten the reason for this should be explored via additional testing. A high test result should not be discounted because the lot mean is below 20 ppm gluten.
- General Mills will continue to use values that fall below the lower limit of quantification for the R7002 assay (i.e., 10 ppm gluten) to determine lot means.
Why this bothers me: Values below the lower limit of quantification are not necessarily accurate. As a result lot means based on values below the LLOQ are not necessarily accurate.
In my opinion, General Mills should switch to the R7001 assay that has a lower limit of quantification of 5 ppm. This would help facilitate accuracy of the lot means.
Update August 11, 2015:
Gluten Free Watchdog Position Statement on Cheerios: Based on the totality of information provided to GFWD, it is our position at this time that individuals with celiac disease should not eat gluten-free Cheerios. Before Gluten Free Watchdog can feel comfortable with this product, General Mills must take steps to ensure that Cheerios are not contaminated with barley (apparently the most problematic grain when it comes to mechanically “cleaning” oats). General Mills must demonstrate via repeated testing of individual boxes of Cheerios using the Ridascreen Gliadin R7001 assay that boxes of Cheerios consistently and reliably test below 20 parts per million of gluten.
Future Discussions with General Mills: General Mills and I have a call scheduled for Thursday, August 13th. General Mills stated to me in email correspondence “We have been working on updates in your 3 areas of concern (oats sourcing, 7002 vs 7001 kit, and composite vs individual lot testing) since your visit, and would like to use this call as a time to update you on some of the progress we’ve made.”
I will continue to update the gluten-free community.
Gluten Free Watchdog is in the process of purchasing Yellow Box Gluten-Free Cheerios for testing.
I recently traveled to General Mills to meet with the people responsible for bringing gluten-free Cheerios to market (Note: General Mills covered most of my travel-related expenses). Cheerios labeled gluten-free will be arriving on supermarket shelves soon (identifiable by the gluten-free statement in the lower right hand corner of the box). The information provided below is based on my visit to General Mills and the testing data shared with me by the company. When gluten-free Cheerios are available, Gluten Free Watchdog will be testing the “yellow box” variety. Once our testing is complete, another statement will be released.
Gluten Free Watchdog current position on oats
Gluten Free Watchdog continues to support the use of gluten-free oats by the celiac disease community that are produced under a gluten-free purity protocol*. The oats used in gluten-free Cheerios are not currently grown under a purity protocol; they are “cleaned” via mechanical sorting. At this time we do not in general support the use of regular commodity oats that are cleaned via mechanical and/or optical sorting to be “gluten-free” after arriving at the mill.
*There are differences in protocols followed by various suppliers of gluten-free oats but they all begin with pure oat seed. Please see the end of this post for a few of the protocols used to produce gluten-free oats.
Data Collection and Test Results for Yellow Box Cheerios Provided by General Mill
General Mills has four plants producing gluten-free Cheerios using what they term “validated gluten-free flour.” At the time of my visit to General Mills (July 14 to 16) they had 88 production dates or “lots” of testing data. Each data point represents a “lot mean.”
General Mills defines a “lot” as a 24-hour production cycle. To arrive at a lot mean, the following protocol is followed:
- Twelve to eighteen boxes of cereal are pulled during a production cycle
- The contents of each individual box are ground
- A sub-sample of ground product is taken from each box
- The sub-samples are composited—meaning they are combined
- The combined sub-samples are subject to additional grinding
- Twelve extractions are taken from this combined, ground sample
- Extractions are tested using the Ridascreen Fast Gliadin (R7002) and cocktail extraction solution
Lot means: According to General Mills lot means for Yellow Box Cheerios (finished product) are between 10 and 13 parts per million of gluten.
Extraction values: Results from randomly pulled data sheets were shared with me. The vast majority of extractions from these data sheets were under 20 parts per million (many were below the lower limit of quantification of 10 parts per million); some extractions were above 20 parts per million (one substantially above).
This section of the post was shared with General Mills prior to publication. They inserted the following statement to the above paragraph “This data represents product produced prior to General Mills applying their ‘Gluten-Free’ labeling to the packaging, having been collected during their ‘validation’ period.”
This is new-to-me information from General Mills. At the time of my visit, I was under the impression that the data being shared was for boxes of Cheerios labeled gluten-free.
My opinion on the test results
In my opinion some of the test results indicate a potential problem (e.g., some extraction values from some lots are above 20 parts per million of gluten). Despite the efforts to mechanically “clean” the oats and the checks and balances in place to make sure wheat and barley have been effectively removed (including manual checking of random draws of oats) some gluten (likely from barley) is ending up in Cheerios.
Based on extraction values shared with me I am not fully confident that every box of Yellow Box Cheerios from the 88 lots produced at the time of my visit to General Mills contains a level of gluten less than 20 parts per million.
Why am I not fully confident? Based on the information shared with me, General Mills is testing final product Cheerios via a mixture of samples taken from 12 to 18 boxes. Any contamination in a composite sample is likely diluted. In other words, IF 11 boxes contain well below 20 parts per million of gluten and only 1 box contains above 20 parts per million of gluten, combining the boxes will cause the contamination from the one box to be “diluted.” It also is impossible to know the source box(s) of contamination.
Dilution can be explained as follows: Say we have 11 glasses of fresh water and 1 glass of salt water. If we sample the salt level of each individual glass of water the level of salt in the glasses of fresh water will be relatively low; the level of salt in the salt water will be relatively high. If we mix all 12 glasses together in a pitcher and measure a sample, the level of salt will be much lower than the glass containing only salt water. While the salt level in a glass of water taken from the pitcher may be acceptable in terms of salt content this does not necessarily suggest that the salt level in all 12 glasses of water is acceptable. You would not want to be the unlucky person choosing the glass of salt water.
In email correspondence I asked General Mills if they were basing their decision on the safety of a “lot” of boxed Cheerios on the lot mean or whether individual extractions above 20 parts per million of gluten raised alarm bells.
Their response: “With all the learning we have done on our journey to bring gluten-free Cheerios to market, we know that looking only at a single data point does not represent the entire picture. As we shared with you, we conduct a minimum of 12 tests on any given sample of Cheerios. In doing so, we look first at the lot mean to give us an initial indication of safety. However, we also look at the confidence interval around the lot mean to tell us about the variability of the individual extractions. It is through a thorough assessment of all the data, we can have confidence around the safety of any given sample or lot.”
According to General Mills, their lab (Medallion Labs) is using the Ridascreen Fast Gliadin R7002 with cocktail extraction to test for gluten levels in finished product Cheerios.
Note: The R7002 assay is not the same assay as the fully validated Ridascreen Gliadin R7001 used by Gluten Free Watchdog.
My opinion on the testing assay
There are several important differences between the R7002 and R7001 assays, including the lower limit of quantification (i.e., the lowest level at which the assay can accurately place a number on the level of gluten). The LLOQ for the R7002 is 10 parts per million of gluten; the LLOQ for the R7001 is 5 parts per million of gluten.
Medallion Labs is reporting precise extraction values when they fall below the LLOQ versus the typical reporting practice of “less than 10 ppm.” Reported values below the LLOQ are not necessarily accurate. While these results are for internal purposes only they are being used to determine a lot mean.
In addition, the R7002 has a 30-minute incubation period compared to the R7001 incubation period of 90 minutes. According to what I’ve been told by the lab used by Gluten Free Watchdog, when the incubation period of an assay decreases, the variability in findings may increase.
General Mills stated the following in email correspondence regarding their use of the R7002 assay when the issues mentioned above (e.g., decreasing incubation times and resulting increases in variability) were presented to them:
“As we stated in our discussion last week, we are aware of the differences in the two kits, including the validation status. At this point we’re confident that we can achieve accurate results by applying the R-Biopharm cocktail solution to our extraction step with the 7002 kit and reviewing our n=12 data sets. The other differences you cite could allow the 7001 kit to achieve the lower limit of quantitation and perhaps lower variability. As we discussed, we will be looking at an internal kit-to-kit comparison, resources permitting. Our experiments show the majority of variability in gluten measurement in oat-based samples, is related to the difficulty in getting a homogenous sample despite our efforts on lab grinding techniques. In the meantime, we are applying the statistical treatment we showed you in order to verify the accuracy of lot means used by the project team to make decisions.”
Suggestions for General Mills
In my opinion General Mills should:
- Continue to move towards sourcing oats from farmers who provide the cleanest oats
- If General Mills is finding wheat relatively easy to remove from oats and barley more problematic then the focus should be on sourcing oats from farms that do not grow barley
- Consider individually testing the 12 to 18 boxes of oats pulled from each lot
- By doing so General Mills may be able to get a better idea of the gluten levels within each box of Cheerios
- Use the fully validated R7001 assay for at least some testing
- This may decrease the variability that may be seen in results
© Copyright July, 22 2015 by Tricia Thompson, MS, RD for Gluten Free Watchdog. All Rights Reserved.
Suppliers (i.e., those dealing directly with the growers) of Oats Grown under a Gluten-Free Protocol (not a complete list)
Current gluten-free protocol confirmed on July 22, 2015:
- Oats are planted only on fields that have not grown glutinous grains for at least two years
- Pure gluten free seed is used for planting
- Fields are hand rogued to eliminate any possible contamination at least three times between planting and harvest
- The grower is responsible for rogueing
- GF Harvest inspectors are in every field of oats we contract
- 3rd party crop inspectors are also utilized
- Oats are harvested and transported in equipment that has been inspected and approved by us for our gluten free production
- The combines are inspected and each grower has been trained on proper cleaning techniques
- Oats are processed in our own dedicated GFCO Certified gluten-free oat facility
- A combination of RidaScreen Elisa R5 and RidaQuick Gluten testing is conducted on all incoming seed stock and after processing
- Independent lab gliadin testing is performed on each batch
Current gluten-free purity protocol confirmed on July 17, 2015:
- Oats are planted only on fields that have not grown gluten grains for 4 years
- Pure gluten free seed is planted
- Fields are hand rogued to eliminate any possible contamination
- The grower is responsible for rogueing
- Each field is inspected by a Montana Gluten Free inspector prior to harvest to assure that the rogueing was thorough and the field is clean
- Oats are harvested and transported in certified clean dedicated equipment
- The combines are inspected and each grower has been trained to clean equipment
- Oats are processed in a dedicated gluten-free facility
- Redundant ELISA R5 testing is conducted before and after processing (since 2012 Montana Gluten-Free also uses the ELISA G12)
Current gluten-free purity protocol confirmed July 21, 2015:
- Select growers attend workshops to learn the requirements for growing gluten-free oats
- Only pure pedigreed seed is used
- Fields used to grow gluten-free oats are
- Required to follow specified crop rotation protocol
- Surrounded by an isolation strip
- All machinery and equipment is either thoroughly cleaned or dedicated
- Oats are processed in a dedicated gluten-free facility
- Oats are tested using R5 ELISA Methodology
Marshall Gluten-Free Milling
Farm certified gluten-free oats should be available by October. Please read about the protocol HERE.