Food Allergen Partnership
In October of 1998, the US Food and Drug Administration (FDA) formed a partnership with both the Minnesota Department of Agriculture (MDA) and the Wisconsin Department of Agriculture, Trade and Consumer Protection (WDATCP). This Partnership was in response to an increase of allergen-related recalls and concerns of controls of cross contamination of undeclared allergen residue. The goals of the Partnership were to eliminate duplicate inspection and sample collections, ensure uniform enforcement, obtain current information on allergen awareness and provide feedback on useful allergy control information and training to the food industry.
While it is generally believed that nearly every food or food ingredient could potentially cause an adverse reaction; eight foods are known to cause ninety percent of severe life-threatening reactions. These are peanut, eggs, milk and milk by products, wheat, tree nuts, soy, fish and shellfish.
Although this Partnership looked at control of all food allergens, the Partnership focused on peanut and egg allergens. Ice cream, bakery and candy manufacturers were selected for coverage. Selection was made randomly of small, medium, and large establishments. Establishments were licensed by the State agencies, conducted interstate commerce and were jointly scheduled in the work plan.
Forty inspections were conducted at Minnesota establishments; three ice cream, 31 bakery and six candy manufacturers. Forty-five inspections were made in Wisconsin; 10 ice cream, 23 bakery and 12 candy manufacturers. Establishment selection was not scientifically significant, but provided an overview of current industry practices. During inspections performed as part of the partnership, egg and peanut finished product surveillance samples were collected for analysis at the State laboratories.
To ensure uniform inspectional approach a specialized questionnaire was developed (Attachment 1). This was used to assess industry practices on critical control points of production. The questionnaire focused on production practices such as handling of rework, labeling and product changeover; viewing these as general mechanisms that relate to the control allergens.
State inspectors and FDA investigators attended an Allergen Workshop in August 1999. Presented during the workshop was an overview of the Partnership goals, food allergen awareness, allergen control strategies, inspection approach, sampling plan and concerns, Neogen peanut ELISA kit demonstration and compliance issues. This training was done to foster a uniform, consistent and targeted inspection approach by state and federal counterparts.
Inspections performed as part of the partnership were routine regulatory inspections. To further uniform application of current good manufacturing practices, inspectors and investigators were standardized during joint initial inspections. Inspections were conducted between September 1999 and March 2000. An introduction letter stating the goals of the Partnership, and a questionnaire that was to assist in industry workshop planning were provided to firm management with the notice of inspection (Attachment 2 & 3).
All inspected establishments received a copy of their inspection report (EIR) and laboratory analytical results. Compliance assistance, such as follow-up inspections, sampling or other enforcement action was taken based on the results of these inspections. Most establishments that had Good Manufacturing Practice (GMP) violations and samples that were positive for undeclared allergen residues were called in by the State and/or federal agency to discuss allergen concerns. Recalls were conducted if deemed appropriate. In the most egregious case some firms received a state issued warning letter. The warning letter explained the laboratory results, and the importance of allergen control. Firms were asked to provide a written allergen control plan. Follow-up inspections were made at violative establishments in July and August 2000.
Industry workshops were conducted to provide feedback on the Partnership findings with a goal of increasing awareness of undeclared food allergen issues. Representatives of the partnering agencies summarized inspection and sample results, and discussed the potential sources of food safety, allergy risks and preventive steps to minimize risk of foodborne illness in food products. Workshops were held in Fond du Lac and Madison, Wisconsin on May 8, 9 respectively and in Maple Grove, Minnesota on May 10, 2000. A binder of useful food allergen control information was provided. Information packets were mailed to inspected establishments that could not attend the workshop.
Allergen Control Plan
Food allergens may become a part of a food through unintended routes. Examples of these are inadequate scheduling to protect against cross-contamination; improper cleaning between different products without analytical verification methods; use of allergen containing rework in non-allergen containing products. A firm should address these critical control points of manufacturing in an allergen control plan.
Nine establishments in Minnesota, which were inspected, had written recall procedures. Only six of their call procedures address allergens. In Wisconsin 20 establishments that were inspected had written recall procedures. Ten establishments’ written recall procedures addressed allergens.
Of the nine Minnesota establishments with written recall procedures, 5 have initiated recalls. Although the MDA and FDA do not classify recalls for product quality, 2 bakery manufacturers had recalled product due to product quality. One bakery and one ice cream manufacturer had recalled product for pathogens, and one ice cream manufacturer recalled a product for undeclared almonds.
Verify Label Accuracy
Our questionnaire contained a section on the use of allergenic ingredients. Although our study focused on egg and peanut allergens the questionnaire also addressed milk proteins, tree nuts, wheat, soybeans, seeds (cotton seed, poppy seeds, sunflower seed) sulfites and yellow #5. Product packaging and labels were verified as correct for the formula being produced. If an ingredient was part of the formulation, but not declared on the label, a sample was obtained for analysis.
During the firm inspection, raw ingredients used in the formulation were compared to the finished product label. Owners of small firms reviewed daily labeling where as in larger firms the production manager or quality assurance may conduct label review. In twenty-five percent of the total establishments inspected there were raw ingredients such as nuts or artificial colors including FD & C Yellow # 5 that were omitted from the label. Manufactured raw ingredients were listed by their common name, but not completely listed on the label. Of the firms that felt they had adequate label verification procedures, 15 percent were found to have label discrepancies. Of the firms which did not verify label accuracy, 38 percent of the Minnesota and 64 percent of Wisconsin firms had product that was analyzed and confirmed positive for undeclared allergen residues.
Figure 1. Summary of Firms Inspected that Verify Labels. Eighty-five bakery, ice cream & candy manufacturers were inspected in Minnesota & Wisconsin. Forty-five have procedures to verify raw ingredient formulation with the finished product label.
Many establishments used private consultants to develop labels. Most firms discard old labels after formulation changes. Other practices utilized by firms to insure label accuracy varied. Some firms would check off ingredients as they were added to the production batch. An ingredient matrix was utilized to schedule product runs. To alert employees of the risk of allergens, some firms put warning labels on bulk allergen ingredients or color-coded bins.
Only one of the 85 firms inspected had a policy against the use of precautionary labeling. One bakery utilized precautionary - "may contain" labeling two years prior to our partnership. Several months ago, they received a consumer complaint from a customer that had an allergic reaction after consuming a product that normally would not contain peanuts. This reaction was apparently due to consumption of peanuts in the product. The customer called the bakery said that they were allergic to peanuts and had a reaction when they consumed the product. The bakery reportedly informed the customer that their labels state the product may contain peanuts. The customer appeared satisfied and the firm has had no additional contact with the complainant. Precautionary labeling may not have been effective because the consumer did not notice.
Figure 2. Comparison of Firms that Verify Labels by Commodity. Twenty-three bakeries, 8 ice cream & 14 candy manufacturers in Minnesota & Wisconsin have procedures to verify raw ingredient formulation with the finished product label.
In an ideal situation there would not be rework or re-feed. Thirty-seven of the 85 firms inspected utilized rework. Of firms that utilized rework, 40 percent of Minnesota and 55 percent of Wisconsin firms had product that tested positive for undeclared allergen residues.
Figure 3. Summary of Firms Inspected that Use Rework. 37 of the 85 bakery, ice cream & candy manufacturers inspected in Minnesota & Wisconsin use rework. Of firms that use rework 40% of MN and 55% of WI had product positive for undeclared allergens.
Figure 4. Comparison of Firms that Utilize Rework by Commodity. Twenty-four bakeries, 4 ice cream & 9 candy manufacturers in Minnesota & Wisconsin have utilize rework in production.
Procedures for handling rework by industry varied. In Minnesota, six of the fifteen firms had documented procedures for handling rework. Three firms used "like-into-like" procedures, two firms did not rework product, which contained nuts, and one firm used color tags, containers and bar coding to document and track how the rework was eventually used.
Procedures to Prevent Cross-Contamination
Product changeover presents an unintentional opportunity for product that contains an allergen to contaminate a product that does not contain that particular allergen, thus resulting in an undeclared allergen. Equipment cleaning is a critical allergen control point for the production of a non-allergen-containing product following product changeover. All visible and non-visible product residues should be removed from product contact surfaces and areas above exposed product zones. Clean up and sanitation steps should be verified and proven effective.
Only four percent of the establishments inspected utilize analytical testing as a method to verify that cleaning and sanitation procedures were proven and effective.
In Minnesota, ten of the 40 firms inspected had Sanitation Standard Operating Procedures that were proven effective and were followed. Samples obtained from these firms did not have detectable allergen residues.
Figure 5. Summary of Firms Inspected with Procedures to Control Cross-Contamination. Forty-one of the 85 bakery, ice cream & candy manufacturers inspected in Minnesota & Wisconsin employed procedures to control cross-contamination.
In many establishments common utensils were used in production of allergen and non-allergen containing products; however three Minnesota firms utilized separate utensils, mixers and equipment to prevent cross contamination. Cross-contamination also occurred when baking sheets were reused without cleaning, or when baking parchments were reused. Several firms reported to reuse baking parchment six to ten times.
Figure 6. Comparison of Firms with Procedues to Control Cross-Contamination by Commodity. Twenty-three bakeries, 13 ice cream & 9 candy manufacturers in Minnesota & Wisconsin have procedures to control cross-contamination.
Production was frequently not scheduled or sequenced for allergen control. Bakeries would schedule production "First-in / First-out" or based on product color. For example, sugar cookies would be the first product manufactured followed by various flavored chip cookies, peanut butter finally ending the production run with gingersnaps. Many firms did not have dedicated equipment for allergen and non-allergen product lines. Non-dedicated product lines were observed to be inadequately cleaned between products, rinsing equipment with water alone or only cleaning equipment at the end of the production day.
One method used to prevent cross contamination of non-allergen and allergen products was scheduling. Production was segregated so that allergen containing products were produced on a separate day of the week than a non-allergen-containing products. Other firms would run peanut containing products at the end of the week or the end of the last production shift. Equipment was completely torn down and cleaned. Cleaning documentation was reviewed and verified prior to the start up of non-allergen products.
Employee awareness and training are the best defenses against unintentional allergen residues being added to products, however only three of the 85 Minnesota and Wisconsin firms inspected utilized personnel that were trained and dedicated to allergen control.
Sample selection was based on inspectional findings. If a product contained an ingredient that was not declared on the label a sample was obtained. If a non-allergen-containing product was produced after an allergen-containing product, a sample was obtained. Or, if a change over had occurred where there was limited or no cleaning between products, a sample was obtained. Samples were obtained from the first production batch following the objectionable observation.
One hundred eighteen samples were collected during the Partnership. Seventy-three samples were analyzed for undeclared peanut protein and forty-five samples were analyzed for undeclared egg protein. Minnesota’s analytical results are comparable to the findings of the Wisconsin establishments. Data presented in the following sections does not include confirmation testing.
Our sampling goal was to assist industry in defining critical control points in their process. Facility deficiencies were identified. If equipment had cracks or pits, or otherwise did not appear easily cleanable, a sample was obtained.
A tolerance for undeclared allergen residue has not been defined. A standard method has not been approved by AOAC. For the Partnership, Minnesota utilized the Veratox® test kit and Wisconsin the Alert® test kit. Both test kits are sandwich enzyme-linked immunosorbent assay (ELISA) developed by the University of Nebraska and manufactured by Neogen Corporation.
The Veratox® kit is quantitative, which was used qualitatively. A 10-ppm kit standard, 10-ppm spike and reagent blank were used as quality assurance controls. Absorbency was read at 630n m with an ELISA reader. No Standard curves were produced for this study. A sample positive for allergen residue was determined based on a 10-ppm, as referenced below.
The Alert® kit is qualitative. The 10-ppm kit standard and a reagent blank were used as controls. Absorbance was checked visually against this standard. Detection was reported as positive at, or above 10 ppm.
Each sample was a composite of 10 eight-ounce sub-samples. Samples were homogenized using a "Robo Coup," and three 500-gram aliquots were reserved (labeled A, B, C). The "A" sample was used for the initial analysis. If the result from this sample was negative, the final result was negative. If the "A" sample was positive, a second chemist preformed the analysis on the "B" sample. If the result from the "B" sample was negative, the final result was negative. If the "B" sample was positive, the final result was determined to be positive. The "C" sample was reserved as the 702(b) portion.
Neogen representative, Patrick Tovey, reported the status of AOAC approval for their peanut allergen test kit. Results from the collaborative study looked very good, but they have asked FDA to independently develop standards and controls for the kit. At that time, they will go through the collaborative process and expect AOAC approval within 12 months.
Minnesota initially used a raw egg to spike a blank sample matrix. From the resulting data, it was clear that it was difficult to obtain consistent results. An all-purpose egg powder was used as a spike. Although the results for this spiking material are more consistent with the 10-ppm standard, the amount of egg present in the all-purpose powder are an approximation (label states a minimum claim). A National Institute of Standards and Technology standard (NIST) may be a more accurate spiking solution.
Egg sample results for this Partnership were compared to the 10-ppm standard for evaluation. Ten percent of samples analyzed were positive for egg protein. Positive products included three ice cream, one donut and 1 cookie sample. Egg allergen, quality assurance data and charts are detailed in Attachments 10 and 11.
Figure 7. Summary of Samples Obtained & Found Positive for Undeclared Egg Protein Residue. Five of the 45 bakery, ice cream & candy samples obtained in Minnesota & Wisconsin were found positive for undeclared egg protein residue in excess of 10 ppm.
Figure 8. Comparison of Samples Obtained for Analysis of Undeclared Egg Protein by Commodity. Two bakery, 3 ice cream & 1 candy samples obtained in Minnesota & Wisconsin were found positive for undeclared egg protein residue in excess of 10 ppm.
Push-through is a common method utilized in ice cream batch change over. Product is run through the line until color changes or designated volume of product has been "pushed through." Some ice cream samples that were found positive for undeclared egg residue were processed following egg-containing product. One positive ice cream sample had egg-containing rework added to the production batch.
One bakery product was found positive. This was a "thumb print" cookie that was cooked for a short period of time. Recent research has shown that the egg protein structure changes during the cooking process and the ELISA kit cannot detect the presence of egg protein, which may lead to false negative results. Subsequent inquiries to Neogen Corporation, Tim Hendra, revealed that the egg protein is denatured during the heating process (cooking or baking) and that the temperature and time cannot be directly correlated to result. The denatured egg protein remains allergenic. It is recommended that the ELISA analysis be performed on raw product, or as a screening tool for validation of cleaning procedures.
A solution made with peanut butter was used to spike a blank sample matrix. The 10-ppm spike and the kit standard had a RPD of 2.766 percent and a range of 0.070 (Attachment 12). It was difficult to bring the peanut butter into solution, which may have contributed to the differences found between the spike and the standard. A standard reference material would be valuable.
Figure 9. Summary of Samples Obtained & Found Positive for Undeclared Peanut Protein Residue. 18 of the 73 bakery, ice cream & candy samples obtained in Minnesota & Wisconsin were found positive for undeclared peanut protein residue in excess of 10 ppm.
Twenty-five percent of the 73 samples analyzed for peanut residue were found positive. Positive products included one ice cream, one donut, eight cookies, two cookie dough and six candy samples. Peanut allergen, quality assurance data and charts are detailed in Attachments 8 and 9.
At one candy manufacturer where a sample was found positive for undeclared peanut residue, fillings were placed by hand onto conveyors and enrobed with chocolate. Enrobers were cleaned once a year and enrobing lines were used for both non-peanut and peanut containing products.
An establishment, which manufactured English toffee and peanut brittle, had a sample that was found positive for peanut residue. This establishment was aware of allergen control strategies, but did not have a written plan. Letters of guarantee were received from suppliers. Ingredients were checked off as added and labels were verified on a batch by batch basis. The practice of rework was not utilized. The firm had dedicated equipment for peanut and non-peanut containing product with the exception of a marble slab table that was utilized in production of all products. This marble table was cleaned between production runs but contained surface knicks.
Figure 10. Comparison of Samples Obtained for Analysis of Undeclared Peanut Protein by Commodity. Eleven bakery, 1 ice cream & 6 candy samples obtained in Minnesota & Wisconsin were found positive for undeclared peanut residue in excess of 10 ppm.
Workshops were conducted to provide current information on food allergens to industry. Agenda topics were determined based on questionnaires that were returned from the inspected establishments. The Allergen Safety Workshop agenda included the following topics (Attachments 13 & 14):
- Human Health Impact
- Industry Impact and Concerns
- Partnership Overview
- Partnership Results
- Inspectional Findings
- Laboratory Findings and Methods Used
- Compliance Follow-up Activities
- Regulations, GMPs, Recalls
- CFSAN Perspective
- Allergen Safety in Industry, and
- Tip in Allergen Safety
Personnel from the targeted industries (chocolate candy, bakeries, and ice cream manufacturers) were invited to attend. State and federal regulators also attended.
Workshops were held in Fond du Lac and Madison, Wisconsin on May 8, 9 respectively and in Maple Grove, Minnesota on May 10, 2000. Workshop evaluations are summarized in attachment 15 through 17. A binder of educational information on food allergen control was provided to all attendees. These binders were also mailed to inspected firms that were unable to attend our workshops.
Follow-up Establishment Inspection
Follow-up inspections were made at establishments that had positive samples for undeclared allergen residues. Nine Minnesota and eleven Wisconsin establishments were inspected in August through September 2000.
State inspectors and FDA investigators attended an Allergen Training Update Workshop during August 2000. Presented during the workshop was a summary of Partnership findings to date, goals for follow-up inspection, and a question and answer period dedicated to allergen control strategies, inspection approach, sampling plan and inspection authority.
The allergen questionnaire was utilized during follow-up inspections. The questionnaire was used to determine if changes were made in the establishment’s general mechanisms to control cross-contamination of allergens. Samples were obtained for analysis of peanut protein. The State laboratories performed the analysis. Summaries of the follow-up establishment inspections are summarized in Attachment 6 for Minnesota and Attachment 7 for Wisconsin.
Establishment Response to Allergen Residue
Many establishments changed operating procedures as a result of inspectional findings, positive sample results, and discussion with the regulatory agency(s). Forty-eight percent of the establishments implemented procedures to prevent cross-contamination of non-allergen and allergen products. Production schedules were developed. Non-allergen-containing products were produced at the start of the manufacturing day, with allergen-containing products produced at the end of the day. Other firms manufactured allergen-containing products on separate days of the week; tree nut products on a separate day than peanut and non-allergen containing products.
Many sanitation changes were made. Equipment was completely torn down, washed with brushes and food grade detergent, rinsed and sanitized. Cleaning was recorded in a log, the log was reviewed and equipment was inspected prior to the start of manufacturing.
Some establishments prevented cross-contamination by dedicating equipment. One firm dedicated a roaster for peanuts and grinders (one for tree nuts, one for peanuts). Many bakery manufacturers dedicated baking pans for peanut products. Baking parchment was discarded after use with a peanut-containing product. Another bakery dedicated a cookie-cutting machine for peanuts only. Utensils were label and color coded for use with non-allergen or allergen products.
Table 4. Summary of Follow-up Inspectional Findings Noting the Firm's Changes in Standard Operating Procedures. Twenty-one follow-up inspections were conducted. Nine Minnesota (7 bakeries, 1 ice cream, 1 candy) and 12 Wisconsin (4 bakeries, 4 ice cream, 4 candy) manufactures were inspected.
Change in Standard Operating Procedures
Employee practices may contribute to allergen residue cross contamination. Two establishments implemented policies where employees were to wash their hands following manufacturing processing that utilized allergen ingredients.
Raw ingredient labels should be verified with the formulation, and ingredients checked as they are added to the production batch. Procedures to verify label accuracy were implemented by 23 percent of the establishments. One firm stopped purchasing artificial almonds that were made from peanuts, replacing the ingredient with raw almonds. Three establishments modified label ingredient statements to declare peanuts instead of generic nuts. Three bakeries discontinued producing peanut butter cookies, and two bakeries stopped grinding and using peanuts on their product.
Of the establishments that used rework, three developed like-into-like procedures and twenty percent discontinued using rework in production. Monitoring finished products for undeclared allergen residues by use of an analytical test method was implemented by 15 % of the firms.
Sample Analysis - Peanut Allergen
Eighteen samples were obtained during the follow-up inspections; seven in Minnesota and eleven in Wisconsin. Two candy samples from Wisconsin were confirmed positive for peanut residue at, or above 10 ppm.
The first positive sample was of English Toffee. The sampled product was manufactured after Milk Chocolate Peanut Clusters. This small candy manufacturer stated during the previous compliance meeting that they could not dedicate a line for peanuts only. Both peanut and non-peanut containing products are manufactured on the same lines/equipment and enrobed in chocolate. All products are now labeled with a precautionary statement "May Contain Peanuts."
The second positive sample was of French Mint Candy manufactured on July 28, 2000. The report states that the last product containing peanuts was produced on July 7, 2000. No additional products were made for 10 days due to equipment breakdown. A sample collected of Almond Rocha Toffee manufactured on July 17, 2000 was negative for peanuts. Both products were sampled on August 8, 2000.
The establishment that manufactured the French mint candy and almond rocha Toffee made several corrections because of the previous allergen partnership inspection. Corrections included: 1) a written allergen and cross-contamination prevention plan; 2), purchase and use a roaster for peanuts only; 3) elimination of reusable pan liners and converting to single use pan liners; and 4) their old nut grinder is used for tree nuts only, and peanuts are ground in a closed food processor. They have now added a statement to their peanut containing products and boxes that state "This container contains a candy with peanuts", instead of only labeling peanuts as "nuts". They have also ordered new labels that state peanuts instead of nuts.
Industry awareness is essential in the control of potential allergen residue risk. Undeclared allergen residues can be introduced during all phases of production. Special emphasis should be placed on assessing raw materials, production scheduling, equipment design, use of rework, sanitation and in packaging and labeling.
An allergen prevention plan must determine potential sources of contaminating allergens and identify appropriate controls to prevent their introduction into food products. These controls may include scheduling production of allergen-containing products at the end of manufacturing runs, appropriate labeling and proper use of rework, equipment and system-design considerations, and thorough cleaning of lines after running allergen containing food products. Proper labeling of food products, effective management of label inventories, control of ingredients from suppliers and training of employees are key factors in allergen controls.
- Wilson Barfknecht
- Andy Cibert
- Steve Clancy
- Marla Hanno
- JoNell O’Neil
- Jan (Stokesbary) Kelly
- Thomas Donaldson
- Karen Jirele
- Kenneth Libertoski
- David Wieneke
- Kristine Zuroski
- Jeff Bernhardt
- Denise Burosh
- Kathy Clarke-Girolamo
- Doug Nelson
- Jon Polzine
- Jill Ball
- Jackie Owens
- Nancy Smit
- Erin Nutter
- Dave Steinhardt
- Karen Di Palma*
- Treeske Ehresmann
- Librada Bautista
- Kathleen Manner
- Jim Selkirk
- Patricia Amundson*
- Tim Anderson
- Lorna Girard
- Theresa Dziuk**
- Cheryl Bigham
- Tony Duran
- Robert Snell
- Richard Willey
- Howard Manresa
- Thomas Nelson
- Michael Barnett
- Laura Murphy
* No longer employed by WDA
** Currently employed by FDA
Minnesota Department of Agriculture
Dairy & Food Inspection Division
90 West Plato Boulevard
Saint Paul, Minnesota 55107-2094
Laboratory Services Division
90 West Plato Boulevard
Saint Paul, Minnesota 55107-2094
US Food & Drug Administration
240 Hennepin Avenue
Minneapolis, Minnesota 55401
Wisconsin Department of Agriculture
Trade and Consumer Protection
2811 Agriculture Drive
Madison, Wisconsin 53708-8911
Bureau of Laboratory Services
4702 University Avenue
Madison, Wisconsin 53707-7883
Letter About the Food Allergen Partnership March 26, 2001