Challenge of Precautionary Allergen Labeling for Ensuring the Safety of Children With Food Allergies

Abstract

Background

Precautionary allergen labeling (PAL) is mandatory and legally regulated in Korea. This study aims to investigate the frequency of PAL use in food products, evaluate its competence, and seek direction for improvement.

Methods

Cow’s milk (CM) and hen’s egg white (EW) protein concentrations were measured using an enzyme-linked immunosorbent assay (ELISA). The results validated PAL using the Voluntary Incidental Trace Allergen Labeling^® 3.0 program. A survey was conducted on guardians to compare preferences and understanding of the current and the arbitrarily revised PAL.

Results

PAL was used in 91.8% (280/305) of baby food products. ELISA results using randomly selected baby food products showed that only 16.7% (5/30; No PAL with no contamination, n = 4; PAL with real-contamination risk, n = 1) were validated to PAL. A detectable CM was found in two products (2/26, 7.7%), with one product exceeding the reference dose (10.3 ± 0.17 ppm). EW was not detected at all (0/16). A total of 207 surveys from guardians were collected and categorized into three groups: food allergy (FA, n = 103), diseases other than food allergies (Others, n = 52), and no disease (Control, n = 52). The FA group exhibited the highest frequency of checking food allergen labeling (“always”: 78.6%, “often”: 9.7%), with a similar PAL adherence (“always”: 58.3%, “often”: 10.4%). None of the groups were satisfied with the current PAL. The ‘allergen-free’ statement was mostly preferred across all groups. The FA group notably preferred PAL with concentration statements.

Conclusion

PAL is excessively prevalent and insufficient in ensuring the safety of children with FAs, necessitating a revision towards a more patient-friendly, evidence-based system for affected individuals and their families.

Graphical Abstract

INTRODUCTION

To date, the standard treatment for food allergies (FAs) is eliminating the culprit foods. Patients must avoid food products that 1) contain the allergen, 2) are processed, cooked, or shared facilities with allergen-containing products, and 3) cross-reacts.1 The FA action plan from the Ministry of Food and Drug Safety includes checking precautionary allergen labelings (PALs) to ensure adherence to the elimination diet. For cow’s milk (CM) and hen’s egg white (EW), which are commonly used as primary or sub-ingredients for various baby food products and account for over 70% of FAs in children under two years old2,3; many are checking PAL regardless of their clinical threshold.4 This can potentially lead to excessive restriction of intake, resulting in nutritional loss and reduced quality of life.5,6

Unlike in Europe and the United States, the PAL in Korean food products is mandatory.7 Since 2021, 22 food allergens have been legally labeled,8 and at the same time, it is legislated that they must be written as PAL unless it is certain that there is no risk of contamination by these allergens, regardless of the amount. In other words, it is safer for the manufacturers to write PAL, “This product is manufactured in the same manufacturing facility as products that contain an anonymous (ANON) food allergen,” to avoid a recall. With PAL becoming more prevalent, it is more likely to be overlooked or neglected on food labels. Consequently, the patients tend to interpret PAL inconsistently, and it fails to meet its original purpose.9,10

The Australian Allergen Bureau established the Voluntary Incidental Trace Allergen Labeling (VITAL^®) program in 2007 by statistically analyzing the reference dose of each food allergen that is not likely to elicit allergic reactions based on the results of several clinical studies.11,12 Action levels on this program is a decision tree expressed as concentrations (ppm) that focused on the amount of protein (mg) in a serving size. The green level (action level 1) indicates that a precautionary cross-contact statement is not required, while the red level (action level 2) indicates that such a statement is necessary.13 Whether VITAL^® can be consistently applied to patients with a severe FA at any age still needs to be evaluated,4,14 but the program is continuously updated11,12,15 and is used by several industries and national food safety agencies.16,17,18

This study aimed to investigate the prevalence of PAL in Korean baby food products and determine the incidence of CM and EW contamination. By synthesizing the survey results on PAL, we aim to understand how PAL can be improved for patients with FAs.

METHODS

Enzyme-linked immunosorbent assay (ELISA) for CM and EW proteins in baby food products with PAL

In August 2020, we conducted an investigation into the PAL of 305 different types of processed baby foods across three major supermarkets in Gyeonggi-do. A total of 10 drinks, 10 snacks, and 10 weaning foods that had a high market share and did not list CM or EW on the raw ingredient labels were randomly selected. An ELISA kit (Veratox^® Total Milk kit, Total Egg kit; Neogen, Lansing, MI, USA) was used for quantitative analysis following manufacturer instructions. The extraction solution was prepared by mixing 10 mM dry powder extraction solvent with 1 L distilled water and preheated at 60°C. Using an electronic scale (XS2002S; Mettler Toledo, Greifensee, Switzerland), 5 g of solid sample finely powdered with a mixer (SMX-W350; Loving Home, Seoul, Korea) or 5 g/mL of liquid sample was weighed in a 250 mL plastic extraction bottle. Samples were mixed with one scoop of the extraction additive and 125 mL of the 60°C extraction solution. The tube was shaken in a shaking incubator (Shaking Water Bath VS-1205SW1; Vision, Daejeon, Korea) at 150 rpm for 15 minutes. After shaking, the samples were collected after centrifugation at 3,000 rpm for 10 minutes (Heraeus Megafuge 16R Centrifuge; Thermo Fisher Scientific, Dreieich, Germany). Then, 100 μL of the standards and sample extracts were simultaneously transferred to the antibody-coated wells and incubated for 10 minutes at room temperature (18–30°C). After 10 times of washing, 100 μL of conjugate was dispensed into each well, and after another wash, 100 μL of the K-blue^® substrate was also dispensed into each well. Next, the red stop solution (100 μL) was dispensed into each well to stop the enzyme reaction. A microplate reader (Micronaut Skan; Merlin Diagnostika GmbH, Bornheim, Germany) was used to measure the absorbance at 620 nm. Measurements were recorded three times, and the mean value was used for the analysis. The ELISA kit’s detection limit was 2.5 ppm, with a detection range of 2.5–25 ppm.

Validation of PAL

PAL was validated according to the VITAL^® 3.0 program.11,15 The reference dose, e.g., eliciting dose for 1% of the patients, for both CM and EW, was 0.2 mg/kg,11,12,15 and multiplied by 1,000/serving dose (g). As a result, if the concentration (ppm) was calculated to be above the action level, PAL was represented as real-contamination. PAL statements were considered ‘validated’ if 1) no PAL with no contamination or 2) PAL with real-contamination risk.

The Action Level (ppm) = Reference Dose (mg) × {1,000/Reference Amount (g)}

Survey on guardians of children

The survey (n = 207) was conducted among guardians of children with a FA (FA group, n = 103), diseases other than food allergies (Other group, n = 52), and no diseases (Control group, n = 52). FAs were confirmed by pediatric allergists, and all patients in FA group were under an elimination diet for culprit foods. The survey assessed the clinical characteristics of children and guardians, as well as preferences for various types of PAL (Appendix 1). The survey consisted of 12 items on basic assessment (two items), FA diagnosis, causative food and anaphylaxis experience (three items), allergen label perception & preferences (two items), understanding, favorability, and reliability of current and different types of arbitrarily revised PAL (five items).

Statistical analysis

Sample size estimation was conducted using the G*Power program. The variables to be considered in calculating the sample size were: a significance level of 0.05, a power of 80%, and an effect size of 0.4. Stata/MP version 15.1 (Stata Corp LLC, College Station, TX, USA) was used for statistical analysis. A χ² test was conducted to determine the frequency distribution among the survey groups. If significant differences were found, post-hoc pairwise comparisons were performed. A Likert scale from 1 point (very low) to 5 points (very high) was used19 to assess reliability, favorability, and understanding according to the type of PAL, and independent samples one-way analysis of variance was conducted for comparison. The results were expressed using mean and standard deviation. P value < 0.05 was considered statistically significant.

Ethics statement

This survey was approved by the Ajou University Hospital Institutional Review Board (IRB), and all subjects were given a copy of the complete Informed Consent Form (AJIRB-SBR-SUR-20-460).

RESULTS

Prevalence and the validation of PAL

Table 1 shows the results of PAL validation for 30 randomly selected products. Among 305 baby food products, 280 (91.8%) had PAL. Among the PAL-labeled products randomly selected for ELISA, CM (2/26; 7.7%) was detected in one drink (No. 7) and one snack (No. 3). EW was not detected (0/16). The CM concentration in drink No. 7 was 10.3 ppm, which was above the action level (0.2 × 1,000/240 = 0.83) and had a real risk of contamination. The CM concentration in snack No. 3 was 2.53 ppm, below the action level (0.2 × 1,000/28 = 7.14). Consequently, 16.67% (5/30; No PAL with no contamination, n = 4; PAL with real-contamination risk, n = 1) of the products were validated to PAL (Table 1).

Table 1. Validating PAL in randomly selected baby food products: on CM and EW.

Product	PAL	CM (ppma)	EW (ppm)	Validationb
Drink 1	Milk, soybean, peanut, wheat, egg, tomato, peach, walnut, chicken	ND	ND	Invalidated
Drink 2	Milk, tomato, peach, soybean, wheat	ND	ND	Invalidated
Drink 3	No PAL	ND	ND	Validated
Drink 4	Milk, egg, peanut, wheat, buckwheat, walnut, soybean, peach, pine nut, tomato	ND	ND	Invalidated
Drink 5	Milk, tomato, soybean, wheat	ND	ND	Invalidated
Drink 6	Tomato, peach, buckwheat, milk, soybean, peanut, wheat, egg	ND	ND	Invalidated
Drink 7	Milk, egg, buckwheat, peanut, wheat, walnut, crab, peach, tomato, sulfite, pine nut	10.3 ± 0.17	ND	Validated
Drink 8	Milk, egg, peanut, wheat, buckwheat, walnut, soybean, peach, pine nut, tomato	ND	ND	Invalidated
Drink 9	Egg, milk, buckwheat, soybean, wheat, mackerel, crab, shrimp, pork, peach, tomato, sulfite	ND	ND	Invalidated
Drink 10	Tomato, egg, peach, wheat, soybean, milk	ND	ND	Invalidated
Snack 1	No PAL	ND	ND	Validated
Snack 2	Milk, peanut, soybean, wheat, shrimp, walnut, chicken	ND	ND	Invalidated
Snack 3	Milk, soybean	2.53 ± 0.29	ND	Invalidated
Snack 4	Milk, peanut, walnut, pine nut	ND	ND	Invalidated
Snack 5	Milk	ND	ND	Invalidated
Snack 6	Egg, milk, buckwheat, peanut, soybean, wheat, mackerel, crab, shrimp, pork. peach, tomato, walnut, chicken, beef, squid, shellfish, pine nut	ND	ND	Invalidated
Snack 7	Milk, soybean, wheat, tomato, walnut	ND	ND	Invalidated
Snack 8	Milk, egg	ND	ND	Invalidated
Snack 9	Egg, milk, soybean, wheat, walnut	ND	ND	Invalidated
Snack 10	Wheat, milk, egg, peanut, soybean	ND	ND	Invalidated
Weaning food 1	Egg, milk, buckwheat, peanut, crab, shrimp, pork, peach, tomato, walnut, chicken, squid, shellfish	ND	ND	Invalidated
Weaning food 2	Egg, milk, buckwheat, soybean, wheat, crab, shrimp, pork, tomato, walnut, chicken, squid, pine nut, shellfish	ND	ND	Invalidated
Weaning food 3	No PAL	ND	ND	Validated
Weaning food 4	Milk, soybean, peanut, wheat, egg, tomato, chicken, pork, shellfish, sulfite	ND	ND	Invalidated
Weaning food 5	Egg, milk, buckwheat, peanut, soybean, mackerel, crab, shrimp, pork, peach, tomato, sulfite, walnut, chicken, beef, squid, shellfish, pine nut	ND	ND	Invalidated
Weaning food 6	Milk, buckwheat, peanut, soybean, wheat, mackerel, crab, shrimp, pork, peach, tomato, walnut, chicken, beef, squid, shellfish	ND	ND	Invalidated
Weaning food 7	Milk, soybean, peanut, walnut, tomato, chicken, peach, shellfish	ND	ND	Invalidated
Weaning food 8	Milk, wheat	ND	ND	Invalidated
Weaning food 9	No PAL	ND	ND	Validated
Weaning food 10	Egg, milk, soybean, wheat, peanut, mackerel, crab, shrimp, peach, pork, chicken, tomato, sulfite, walnut, squid, shellfish, pine nut	ND	ND	Invalidated

PAL = precautionary allergen labeling, CM = cow’s milk protein concentration, EW = egg white protein concentration, ND = not detected.

^appm: mg/kg.

^bBy using Voluntary Incidental Trace Allergen Labeling^® 3.0.

Characteristics of survey respondents’ children with FA

The most common age group for patients in FA group was ≥ 6 years (38.8%, n = 40), followed by 36 months to < 6 years (36.9%, n = 38), 12 months to < 36 months (23.3%, n = 24), and < 6 months (1.0%, n = 1). The frequency of allergenic foods in the FA group, in decreasing order, was as follows: EW (n = 73), CM (n = 52), peanut (n = 36), wheat (n = 33), walnuts (n = 28), buckwheat (n = 23), pine nuts (n = 17), shrimp (n = 16), crab (n = 15), mackerel (n = 11), soybean (n = 10), peach (n = 8), shellfish (n = 8), tomato (n = 4), and others (n = 37). Of the 103 guardians in the FA group, 80 (77.7%) reported having children allergic to two or more foods, with 57 (55.3%) reporting that their children had experienced anaphylaxis.

Awareness of food allergen labeling and PAL

One guardian per patient was surveyed and the majority of were in the 31–40 years age group (65.7%, 136/207), followed by the 41–50 years age group (30.9%, 64/207), with smaller percentages for the 51–60 years age group (2.4%, 5/207) and the 21–30 years age group (0.5%, 1/207). In the FA group, the majority showed high awareness of food allergen labeling, with 78.6% always aware, 9.7% often aware, 4.9% sometimes aware, 3.9% rarely aware, and 2.9% never aware. In contrast, in the Others group, there was a notable trend of not checking food allergen labeling, with 34.6% never checking, 26.9% rarely checking, 13.5% sometimes checking, 13.5% often checking, and 11.5% always checking. Compared to the Others and Control groups, the FA group checked food allergen labeling more often (P < 0.001). In the FA group, 78.6% of guardians always or often checked PAL, with 58.3% always checking and 20.4% often checking. In contrast, the Others and Control groups had the highest number of guardians who never checked PAL. The FA group demonstrated significantly higher PAL checking frequency compared to both the Others (34.6%) and Control (32.7%) groups (P < 0.001; Table 2).

Table 2. Awareness of food allergen labeling and PAL among survey respondents.

Variables		Total (n = 207)	FA group (n = 103)	Others group (n = 52)	Control group (n = 52)	χ2	P
Food allergen labeling						100.36	< 0.001
	Never	36 (17.4)	3 (2.9)	18 (34.6)	15 (28.8)
	Rarely	36 (17.4)	4 (3.9)	14 (26.9)	18 (34.6)
	Sometimes	15 (7.2)	5 (4.9)	7 (13.5)	3 (5.8)
	Often	25 (12.1)	10 (9.7)	7 (13.5)	8 (15.4)
	Always	95 (45.9)	81 (78.6)	6 (11.5)	8 (15.4)
PAL						74.89	< 0.001
	Never	38 (18.4)	3 (2.9)	18 (34.6)	17 (32.7)
	Rarely	38 (18.4)	10 (9.7)	13 (25.0)	15 (28.8)
	Sometimes	26 (12.6)	9 (8.7)	10 (19.2)	7 (13.5)
	Often	34 (16.4)	21 (20.4)	7 (13.5)	6 (11.5)
	Always	71 (34.3)	60 (58.3)	4 (7.7)	7 (13.5)

The survey was conducted among guardians of children with food allergies (FA group), guardians of children with disease other than food allergy (Others group), and guardians of healthy children (no disease; Control group). Values are presented as number (%).

PAL = precautionary allergen labeling.

Preferences for current with revised PAL

Respondents in a survey on five PAL types showed the lowest reliability, favorability, and understanding of the current PAL statement. In particular, the FA group had the lowest reliability (3.35 ± 0.94), favorability (3.26 ± 0.99), and understanding (3.42 ± 1.03) for the current PAL. There was a significant difference in preferences for “Produced in a factory where ANON is used and 1 mg/kg ANON is detected” between the groups. In the FA group, this statement achieved the highest scores among the five PAL types for reliability (4.27 ± 0.97), favorability (4.13 ± 1.04), and understanding (3.94 ± 1.24). In contrast, the Others and Control groups exhibited consistently lower scores across all perspectives than the FA group (P < 0.001). Notably, the Control group scored lower in understanding, with a mean of 2.85 ± 1.18, compared to the current PAL score of 3.29 ± 1.02. An intergroup comparison of preference for “When food allergen is not detected, remove PAL” also varied significantly based on the presence of a FA. This statement received considerably lower scores in reliability and favorability in the Others and Control groups compared to the FA group. Remarkably, within the FA group, this statement outperformed the current PAL in all perspectives. The statement “ANON-FREE/ANON not detected (when an allergen is not detected)” had consistently high scores in all perspectives across all groups. Similarly, all groups scored high in understanding the statement, “When ANON is detected, label it as ‘Contains ANON’” (Fig. 1).

Fig. 1. Preferences for current and revised versions of PAL statements.

The survey was conducted among guardians of children with food allergies (FA group), guardians of children with disease other than food allergy (Others group), and guardians of healthy children (no disease; Control group).

PAL = precautionary allergen labeling, ANON = anonymous.

DISCUSSION

Challenges of PAL have been reported in several studies.18,20,21 However, studies on patient-centered consequences have rarely been conducted,22,23 and those focused on baby food products have not been reported. Herein, we observed that PAL was present in 91.8% of baby food products from three large supermarkets in Gyeonggi-do, Korea, far exceeding that reported in previous studies in other countries.24 Although the PAL rate of all distributed foods has not been investigated in this study, assessing the PAL from baby food products is essential, as FAs are more prevalent in children.25

Since this study was conducted at a single tertiary institution, the findings cannot be generalized to all patients. However, 77.7% of the subjects had multiple FAs of two or more, suggesting that many children with FAs eliminate various foods. There has been a rise in recommending individualized diets for patients with FAs, involving assessments of symptom-eliciting doses and suggestions for a thermal-treated hypoallergenic diet where feasible.26 However, the availability of tailored diets is currently confined to specialized allergy clinics.25 It remains indisputable that the general action plan for patients is eliminating not only food products containing allergens but also those posing a risk of contamination.1,3 In this scenario, strict adherence to a comprehensive elimination diet, including all indications on PAL, could considerably restrict the diet options for children with FA, given the prevalent use of PAL.

The present study showed that products with no real risk excessively indicate PAL. Interpretation needs to be cautious since, in the VITAL^® program, the reference dose is now mainly based on the results from oral food challenge studies. There may be limitations in protecting infants under the age of two years or patients with anaphylaxis who rarely undergo oral food challenge tests. Also, the range of food proteins that demand various conversion factors can be problematic in detection methods, including ELISA.27 However, the PAL validation result from this study only covers 16.7% of baby food products, which clearly emphasizes the need to reconsider PAL and its regulation.

The FA group consistently demonstrated heightened vigilance in checking PAL, surpassing participants in other studies. Despite PAL’s low likelihood of indicating real-contamination risk, a significant number in the FA group consistently reported checking PAL often or always. This contrasts starkly with prior U.S. findings, where only 36% believed in the accuracy of PAL, and 19% admitted to ignoring it.6 While the age of the children may contribute to this variance, it is plausible that uniform guidelines emphasizing PAL verification have influenced these results. To ensure the efforts of children’s guardians are not in vain, PAL must precisely reflect the real risks of allergic reactions.

We arbitrarily developed four types of PAL based on observations from other countries. Although PAL was infrequently checked in the Others and Control groups, guardians of children in these groups expressed reluctance to delete PAL when the food product was not contaminated. Conversely, the FA group preferred PAL only when there was actual contamination. The FA group are relatively fewer in number, yet indicating PAL on products should serve the explicit purpose of protecting those suffering from FAs. Consequently, the findings of this study underscore the necessity for revising PAL. The PAL statements “Contains ANON” and “ANON-FREE/ANON not detected,” now widely adopted globally, demonstrated high understanding, favorability, and reliability irrespective of children’s diseases or FA. This implies a favorable potential for PAL modification.

This study utilized the VITAL^® program, established 17 years ago and continuously evolving, to standardize cross-contact statements. This program helps manufacturers transition from probability-based PAL to concentration-based verification. The ELISA assay method used in the study was an easy and cost-effective means of assessing the need for PAL at the VITAL^® level. Based on our findings, we suggest that, rather than relying on the concept that PAL ‘may not be zero,’ actual measurement efforts should be made to indicate when allergen concentrations are significant. Efforts to improve PAL, as described above, can provide valuable data for determining the symptom-eliciting dose or threshold for patients in the future. In summary, the PAL for baby food products in Korea does not adequately reflect the real contamination risk and is overused. A comprehensive review is necessary to revise PAL into a more patient-friendly, evidence-based system that ensures better protection for patients with FAs and their families.

Appendix 1

Survey questionnaire