Comparison of Double‐Blind and Open Food Challenges for the Diagnosis of Food Allergy in Childhood: The ALDORADO Study

ABSTRACT

Background

Double‐blind placebo‐controlled food challenge (DBPCFC) is widely regarded as the “gold standard” to diagnose food allergy. Maximum efforts are made to reduce bias, yet DBPCFCs are costly, time‐, and resource‐intensive. Less demanding open food challenges are increasingly used in clinical practice. However, recommendations regarding the use of these challenges are based on low certainty of evidence, and no comparative studies have been performed using the most recent international food challenge guidelines. We hypothesised that the open food challenge is non‐inferior to DBPCFC in children suspected of allergy to cashew nuts, hazelnuts or peanuts.

Methods

A total of 63 children, aged 4 years and older, were included if referred for suspected IgE‐mediated allergy to cashew nut, hazelnut, or peanut. All study participants underwent DBPCFC first, followed by an open food challenge for the same food. Challenge outcomes were assessed by predefined criteria into positive, negative, or inconclusive.

Results

DBPCFC and open food challenge outcomes were the same for 36/41 (87.8%) patients. Sensitivity and specificity of the open food challenge were 0.91 (95% CI 0.79, 1.03) and 0.83 (95% CI 0.63, 1.01), respectively, with an AUC value of 0.87. Eliciting and stop doses were not significantly different between both food challenges.

Conclusion

The Diagnostic accuracy of open food challenge is non‐inferior to that of DBPCFC. This finding implies less demanding open food challenges can be implemented for children from the age of 4 years suspected to be cashew nut, hazelnut, or peanut allergic. Further research is necessary to validate our findings and to investigate the diagnostic accuracy for other major food allergens.

We performed the first non‐inferiority diagnostic study comparing food challenge outcomes. The Open food challenge outcome was found to be non‐inferior compared to double‐blind placebo‐controlled food challenges for children aged 4 years and older suspected of cashew nut, hazelnut, or peanut allergy. Our results imply that less demanding open food challenges may be implemented for selected children in clinical practice.Abbreviations: AUC, area under the curve; DBPCFC, double‐blind placebo‐controlled food challenge; NPV, negative predictive value; PPV, positive predictive value.

1. Introduction

The prevalence of confirmed food allergies increased from 2.6% (2000–2012) to 3.5% (2012–2021) within European regions [1]. An accurate diagnosis of food‐allergy is important, as food allergic patients can be educated to avoid the allergen, use an epinephrine auto‐injector, or be selected for immunotherapy, whereas excluding the diagnosis may result in (re)introduction of the allergen in the diet and reduction of anxiety [2, 3].

Current guidelines support the use of oral food challenges (OFCs) to diagnose food allergies [4, 5]. The double‐blind placebo‐controlled food challenge (DBPCFC) is widely regarded as the “gold standard” as maximum efforts are made to reduce bias, yet this is a costly, time‐, and resource‐intensive procedure [4, 6]. In contrast, open food challenges may be more susceptible to false‐positive outcomes, as supported by the occurrence of allergic symptoms on placebo days of DBPCFCs, with previously reported prevalence varying from 2.8% to 12.9% [7, 8, 9]. These symptoms are presumably due to stress on the challenge day, expectation bias, and fluctuation of (allergic) symptoms that are attributed to the tested allergen [10].

Open food challenges are increasingly used in clinical practice for practical reasons, especially if objective outcomes are expected [5]. However, the recommendation to conduct an open challenge or DBPCFC is based on low certainty of evidence. Only two prospective studies, both published in 2007, have been performed to compare the outcome of both challenges. The first trial included 41 children suspected to be allergic to cow's milk, egg, wheat, and prawn. Patients underwent an open food challenge, followed by DBPCFC (in some cases performed at home) only if the open food challenge outcome was positive. According to results of this trial, open food challenges were recommended in case of immediate objective symptoms, and in the event of delayed (mainly subjective) symptoms, DBPCFCs were indicated [11]. The second study included 20 adults suspected to be allergic to multiple organic foods. Only one of the foods that induced a reaction during the open food challenge was selected for DBPCFC. Observed reactions were mostly classified as non‐IgE mediated. The authors concluded that, regarding confirmation of food allergy, open food challenges are comparable with DBPCFCs [12]. In both studies, no standardised procedures were followed for conducting OFCs, and no internationally validated symptom scoring system was applied [11, 12]. To date, no comparative study has been reported based on the outcomes of DBPCFC and open food challenges using the most recent practice guidelines for performing OFCs. According to current guidelines of international paediatric allergists, there is an urgent need to investigate the diagnostic accuracy of open and double‐blind food challenges for individual food allergens [5].

Therefore, we aimed to compare the outcome of DBPCFCs and open challenges. Within this first non‐inferiority diagnostics study, the ALlergy Diagnosed by Open oR DOuble blind food challenge (ALDORADO) trial, we hypothesised that the diagnostic value of open food challenge is non‐inferior to the “gold standard” DBPCFC in children with suspected allergy for cashew nut, hazelnut, or peanut.

2. Methods

2.1. Study Population

The ALDORADO trial was a single‐centre non‐inferiority study conducted at a paediatric allergy clinic of a large teaching hospital in the Netherlands from October 2021 until February 2024. Patients aged four to 18 years were eligible for inclusion if referred for suspected IgE‐mediated allergy to cashew nut, hazelnut, or peanut; and if an OFC was advised after thorough medical and dietary assessment. Patients who were re‐evaluated for their food allergy status were also eligible for inclusion. Patients were excluded if they used beta blockers and/or prednisolone, suffered from uncontrolled asthma, had congenital pulmonary or cardiac comorbidity, were pregnant, or were unable to adequately report the occurrence of possible symptoms. Sensitisation to the suspected food allergen was defined as serum‐specific IgE (sIgE) level ≥ 0.35 kU/L (Phadia 250, Uppsala, Sweden) [13]. sIgE was determined in our laboratory, unless this had already been done within 6 months before inclusion. The study was approved by the Dutch Central Committee on Research Involving Human Subjects (CCMO; NL76237.000.21) as well as the Institutional Review Board (MEC 2021‐039) and was registered as a clinical trial in the Dutch Clinical Trial Registry (URL: https://onderzoekmetmensen.nl/en/trial/27959). This report is in adherence with the Standards for Reporting of Diagnostic Accuracy Studies (STARD) guidelines (Table S1) [14]. We previously published the rationale and study design for this trial, for which parents of (suspected) food allergic children provided their input [10]. Informed consent was obtained from parents of children aged up to 12 years, from parents and children aged 12–16 years, and from children alone aged 16 years and older.

2.2. Oral Food Challenges

For all study participants, both the DBPCFC and open food challenge were scheduled for the suspected food allergen. DBPCFC consisted of two separate days: one active day (i.e., administration of suspected food allergen) and one placebo day. Both days were scheduled in random order. The open food challenge consisted of one day, on which the suspected food allergen was administered. DBPCFC was always scheduled first. Both challenges were planned with an interval ranging between 1 to 6 weeks. All challenges were performed according to European Academy of Allergy and Clinical Immunology (EAACI) guidelines [4]. During both challenges, a cumulative dose of 4443 mg of allergenic food protein was administered, starting at 3 mg and divided into a maximum of seven steps (increasing dose), with a 30 min observation time between each dose. For the open food challenge, the validated recipe for the active day of DBPCFC was used to eliminate potential matrix effects on the challenge outcome [15]. Symptoms were registered using the scoring system presented by Grabenhenrich et al. [16] (Table S2). All challenge outcomes were discussed by the research team (paediatric allergist, paediatrician, and medical researcher) and assessed according to predefined criteria into positive, negative, or inconclusive as shown in Figure S1 [10]. If mild symptoms (e.g., itchy mouth or abdominal pain) occurred and did not aggravate in the next 30 min, patients were encouraged to eat the same dose again to determine OFC outcome. Safety was guaranteed by ending the specific OFC if severe allergic symptoms as shown in Table S3 occurred [10]. In case other symptoms occurred, OFC was ended if clinically indicated. After administration of the final dose, patients were observed for at least 2 h. Parents were instructed to report possible symptoms at home within 24 h after the challenge was ended. Results of DBPCFC were reviewed before the second challenge was performed. In case of an anaphylactic reaction during DBPCFC, no open food challenge was performed due to safety reasons [17]. DBPCFC outcomes were not reported to the patient until after the open food challenge. Challenge days were planned in such a way that the second challenge was supervised by another trained paediatric nurse, who was not aware of the results of the first challenge. Doctors, of whom nurses required assistance in case of more severe reactions, were blinded too.

2.3. Outcomes

The primary endpoint was the difference in the proportion of positive outcomes of DBPCFC and the open food challenge. Open food challenges from children who were not willing to perform the open food challenge after DBPCFC was performed or who were deemed unsafe due to anaphylaxis during DBPCFC were excluded in the main analysis and defined as positive in the sensitivity analysis as predefined in the research protocol, respectively. Secondary endpoints included the difference between DBPCFC and open food challenge in symptom severity, eliciting dose (ED, first dose that causes allergic symptoms), and stop dose (SD, cumulative total dose that has been eaten). OFCs with inconclusive outcomes were excluded from analysis.

2.4. Sample Size

Retrospective data analysis of DBPCFCs for peanuts and tree nuts of our allergy centre showed that 47% were positive, and reaction on placebo days occurred in 17% of all these DBPCFCs [Wouter W. de Weger (W.W.W.) and Arvid W.A. Kamps (A.W.A.K.)], which is consistent with the approximately 15% reported in literature [7]. This percentage of placebo reactions in DBPCFC was interpreted as the expected percentage of additional open food challenges with positive outcomes. Using a non‐inferiority design, we therefore defined the non‐inferiority margin at 15% (H₀: the prevalence of open food challenge with a positive outcome will be equal to or less than 15% higher than the prevalence of DBPCFC with a positive outcome; H₁: the prevalence of open food challenge with a positive outcome will be more than 15% higher than the prevalence of DBPCFC with a positive outcome). Sample size analysis was performed for the comparison of DBPCFC with open food challenges using a power of 80%, a one‐sided α of 0.05, an equivalence difference of 0.15, and an expected proportion of DBPCFC with a positive outcome of 0.47. We performed one‐sided power analysis, because we did not expect an open food challenge outcome to be negative following a DBPCFC with positive outcome. This resulted in a minimum of 60 children that needed to be included (power analysis of non‐inferiority tests of correlated proportions).

2.5. Statistical Analysis

The McNemar test for correlated proportions was performed on the results of both the DBPCFC and open food challenge outcomes. Sensitivity and specificity (including ROC curve) of the open food challenge were determined using DBPCFC as a reference. Sensitivity analysis was performed excluding the not performed open food challenges or including these as positive, respectively. Possible differences between patients with positive and negative DBPCFC outcomes in their basic characteristics were analysed using a two‐sample t‐test (if normally distributed) or Mann–Whitney U test (if not normally distributed). The Wilcoxon signed‐rank test was used for possible differences between eliciting and stopping doses for DBPCFCs and open food challenges. p < 0.05 were considered statistically significant. IBM SPSS Statistics version 25 (IBM, NY, USA) was used for all analyses.

3. Results

A total of 63 patients were included, of whom characteristics are presented in Table 1, stratified by DBPCFC outcome. Patients with positive DBPCFC outcomes suffered more often from a proven allergy for another food, compared to patients with negative DBPCFC outcomes (p = 0.008). DBPCFC outcomes were positive in 33/63 (52.4%), negative in 22/63 (34.9%), and inconclusive in 8/63 (12.7%) cases (Figure 1). Open food challenges were successively performed by 53/63 (84.1%) patients. Open food challenges were not performed in 4/10 cases due to anaphylaxis during DBPCFC and in 6/10 cases per the request of study participants. Open food challenge outcomes were positive in 28/53 (52.8%), negative in 18/53 (34.0%), and inconclusive in 7/53 (13.2%) cases. The mean period between both performed OFCs was 5.4 weeks (standard deviation 3.6 weeks). The planned interval of 6 weeks was extended for 11/53 patients (see Table S4 for reasons).

TABLE 1.

Patient characteristics.

	Total	Positive	Negative	Inconclusive	p
N	63	33 (52.4%)	22 (34.9%)	8 (12.7%)	0.769a
Age (years)	10.0 (4.0–17.0)	10.0 (4.0–16.0)	10.0 (4.0–17.0)	8.0 (4.0–12.0)
Gender
Female	23 (36.5%)	11 (33.3%)	9 (40.9%)	3 (37.5%)	0.567b
Male	40 (63.5%)	22 (66.7%)	13 (59.1%)	5 (62.5%)
Food
Cashew nut	11 (17.5%)	9 (27.3%)	2 (9.1%)	0 (0.0%)
Hazelnut	30 (47.6%)	13 (39.4%)	12 (54.5%)	5 (62.5%)
Peanut	22 (34.9%)	11 (33.3%)	8 (36.4%)	3 (37.5%)
Atopic comorbidities
Allergic rhinitis	50 (79.4%)	25 (75.8%)	19 (86.4%)	6 (75.0%)	0.495c
Asthma	27 (42.9%)	16 (48.5%)	10 (45.5%)	1 (12.5%)	0.825b
Eczema	45 (71.4%)	21 (63.6%)	19 (86.4%)	5 (62.5%)	0.064b
Food allergy	30 (47.6%)	21 (63.6%)	6 (27.3%)	3 (37.5%)	0.008b
Atopic disease (family)
None	6 (9.5%)	2 (6.1%)	4 (18.2%)	0 (0.0%)	0.204c
≥ 1 family member	57 (90.5%)	31 (93.9%)	18 (81.8%)	8 (100.0%)
sIgE for specific food
< 0.35 kU/L †	1 (1.6%)	0 (0.0%)	1 (3.8%)	0 (0.0%)	0.400c
≥ 0.35 kU/L	62 (98.4%)	33 (100.0%)	21 (95.5%)	8 (100.0%)
cashew nut (kU/L)	1.10 (0.37–79.8)	1.60 (0.38–79.8)	0.61 (0.37–0.84)	—
hazelnut (kU/L)	19.0 (< 0.35–> 100)	19.3 (0.45–> 100)	10.4 (< 0.35–> 100)	36 (3.8–76.9)
peanut (kU/L)	1.70 (0.47–48.0)	4.4 (0.59–48.0)	1.05 (0.47–4.7)	1.8 (0.84–2.8)
Previous exposure ‡
None	15 (23.8%)	6 (18.2%)	7 (31.8%)	2 (25.0%)
History in daily life	33 (52.4%)	15 (45.5%)	12 (54.5%)	6 (75.0%)
OFC	15 (23.8%)	12 (36.4%)	3 (13.6%)	0 (0.0%)
Open food challenge
Positive	28 (44.4%)	21 (63.6%)	3 (13.6%)	4 (50.0%)
Negative	18 (28.6%)	2 (6.1%)	15 (68.2%)	1 (12.5%)
Inconclusive	7 (11.1%)	1 (3.0%)	3 (13.6%)	3 (37.5%)
Not performed	10 (15.9%)	9 (27.3%)	1 (4.5%)	0 (0.0%)
Anaphylaxis	4 (40.0%)	4 (44.4%)	0 (0.0%)	0 (0.0%)
Request participant §	6 (60.0%)	5 (55.6%)	1 (100.0%)	0 (0.0%)
Symptoms placebo day	4 (6.3%)	0 (0.0%)	1 (4.5%)	3 (37.5%)

Note: Data stratified for DBPCFC outcome. Results presented as mean ± standard deviation (normally distributed) or as median and min‐max (not normally distributed). Other parameters are presented as frequencies and percentages. Differences between the groups, for DBPCFCs with positive and negative outcomes respectively, were tested with the Mann–Whitney U test (a), Chi‐square (b), or Fisher's Exact (c).

Abbreviations: DBPCFC, double‐blind placebo‐controlled food challenge; sIgE: serum‐specific IgE.

^† No medical reason to perform an OFC, however child/parents preferred OFC due to previous reaction at the suspected food allergen during introduction at home.

^‡ Previous exposure to the same food, categorised as: absent (none); reaction at home, which was mentioned during thorough medical and dietary assessment (history in daily life); re‐evaluated for food allergy status after previous OFC for the same food (OFC).

^§ 5 due to severity of symptoms during DBPCFC; 1 due to time investment.

FIGURE 1.

Flowchart. DBPCFC, Double‐blind placebo‐controlled food challenge; OFC, Oral food challenge.

The outcome of OFCs was the same for 36/41 (87.8%) patients (Figure 2). Of 23 DBPCFCs with a positive outcome, 21 open food challenges were positive, and for 18 DBPCFCs with a negative outcome, 15 open food challenges were negative. In three patients with DBPCFCs with a negative outcome, the open food challenge outcome was positive. Finally, in two patients with DBPCFCs with a positive outcome, the open food challenge outcome was negative. One of these patients suffered from oral allergy symptoms, and the other one experienced occasional sniffing with minimal lip oedema for a short period of time.

FIGURE 2.

Diagnostic performance of open food challenge. All open food challenges that were not performed were excluded. (A) Sensitivity and specificity of the open food challenge were 0.91 (95% CI 0.79, 1.03) and 0.83 (95% CI 0.63, 1.01), respectively. (B) Receiver Operator Curve (ROC) of an open food challenge. The Area Under the Curve (AUC) = 0.87. DBPCFC, Double‐blind placebo‐controlled food challenge.

Symptoms on placebo days occurred during 4/63 (6.3%) DBPCFCs, with negative (1/4) and inconclusive (3/4) outcomes respectively (Table S5). DBPCFC outcome was negative because symptoms only occurred on placebo day and were absent after active day (related open food challenge outcome was also negative). For two DBPCFCs, symptoms occurred during both challenge days, so their outcome was inconclusive (the related open food challenge outcome was positive). For one DBPCFC, symptoms occurred on the placebo day, and the total amount of allergenic protein was not administered completely (the related open food challenge outcome was negative).

Diagnostic performance of the open food challenge was non‐inferior to DBPCFC (Figure 2). Open food challenges performed similarly to DBPCFC for diagnosing food allergies (p = 1.000). Using DBPCFC results as the “golden standard,” the sensitivity of the open food challenge was 0.91 (95% CI 0.79, 1.03), and specificity was 0.83 (95% CI 0.63, 1.01). The calculated AUC value was 0.87. The odds for a positive DBPCFC outcome if the open food challenge outcome is positive are 52.5 (p < 0.0001). Comparable results (AUC = 0.86) were found after defining the not performed open food challenges as positive (Figure S2). Exploratory post hoc analysis indicates that diagnostic performance of the hazelnut challenge was lower than cashew nut and peanut, although this analysis was not part of our initial study protocol (Table S6).

No differences were found between either ED and SD of open food challenges with positive outcome, and related DBPCFCs (with either positive or negative outcome). A total of 24 food challenges were included for analysis, and both stop dose (p = 0.177) and eliciting dose (p = 0.615) were not significantly different between the two OFCs (Table 2).

TABLE 2.

Overview of eliciting and stop dose.

	Food	ED		SD
		Open	DBPCFC	Open	DBPCFC
1	Peanut	100	3000	4443	4443
2	Cashew nut	30		4443	4443
3	Cashew nut	3	3	43	73
4	Peanut	3	3	4443	4443
5	Hazelnut	300	1000	3449	4433
6	Peanut	300	100	443	4443
7	Cashew nut	3	3	43	743
8	Hazelnut	100	100	1443	443
9	Cashew nut	100	3	843	106
10	Hazelnut	10		73	4443
11	Peanut	3	300	6	343
12	Cashew nut	30	10	143	243
13	Hazelnut	3	3	13	173
14	Peanut	3	3	6	13
15	Cashew nut	30	3	143	73
16	Peanut	10	100	4443	1443
17	Cashew nut	1000	30	1443	2443
18	Peanut	3	1000	1943	2443
19	Peanut	10	10	73	73
20	Peanut	3		1443	4443
21	Hazelnut	3	10	443	1443
22	Hazelnut	10	3	4443	4443
23	Cashew nut	2700	10	4143	743
24	Hazelnut	3	1000	1443	1443

Note: Comparison of open food challenges with positive outcomes and related DBPCFCs (active days) with either positive or negative outcomes (n = 24). ED is absent in case DBPCFC outcome is negative.

Abbreviations: DBPCFC, double‐blind placebo‐controlled food challenge; ED, eliciting dose; SD, stop dose.

Symptoms that occurred during OFCs with positive outcomes were moderate in the majority of open food challenges and severe in most DBPCFCs (Table S7). A Detailed overview of all 33 DBPCFCs with positive outcomes and related open food challenges is shown in Table 3 and Table S8. Table S9 shows details of the remaining 7/28 open food challenges with positive outcomes and related DBPCFCs. Anaphylaxis occurred during four DBPCFCs and was absent in open food challenges. Solitary subjective symptoms (i.e., abdominal pain or itchy mouth) were reported during 4/33 (12.1%) DBPCFCs and 13/28 (46.4%) open food challenges with positive outcomes respectively. Of the latter, 8/13 open food challenges followed after DBPCFCs with positive outcomes. The remaining five open food challenges were performed after DBPCFCs with negative (n = 3) and inconclusive (n = 2) outcomes.

TABLE 3.

Overview of stop dose, symptoms (total) and dietary advice.

Overview of stop dose, symptoms (total) and dietary advice for peanut
(A)	Food	SD		Symptoms (total)		Dietary advice
		DBPCFC	Open	DBPCFC	Open	DBPCFC	Open
1	Peanut	3R	3R	S	S	E	E
2	Peanut	4	2	B	S	E	E
3	Peanut	7	5	B	S	E	E
4	Peanut	7	7	B	B	E	E
5	Peanut	7	7	B	B	E	E
6	Peanut	6R	6R	B	B	E	E
7	Peanut	2	1R	B	B	E	E
8	Peanut	6	7	B	B	E	E
9	Peanut	7	ANA	O	ANA	E	ANA
10	Peanut	4	NP	B	NP	E	NP
11	Peanut	5R	NP	B	NP	E	NP

Overview of stop dose, symptoms (total) and dietary advice for cashew nut
(B)	Food	SD		Symptoms (total)		Dietary advice
		DBPCFC	Open	DBPCFC	Open	DBPCFC	Open
1	Cashew nut	3	3	B	S	E	E
2	Cashew nut	6R	6	S	S	E	E
3	Cashew nut	5R	7	S	S	E	E
4	Cashew nut	4R	4	B	B	E	E
5	Cashew nut	3R	4	B	B	E	E
6	Cashew nut	4	6	O	B	E	E
7	Cashew nut	5R	3	B	B	E	E
8	Cashew nut	4	NP	O	NP	E	NP
9	Cashew nut	5	NP	B	NP	E	NP

Overview of stop dose, symptoms (total) and dietary advice for hazelnut
(C)	Food	SD		Symptoms (total)		Dietary advice
		DBPCFC	Open	DBPCFC	Open	DBPCFC	Open
1	Hazelnut	7	7	B	S	R	R
2	Hazelnut	6	5	B	S	E	E
3	Hazelnut	7	7	B	B	R	R
4	Hazelnut	6	6	O	B	E	E
5	Hazelnut	5	6	B	B	E	E
6	Hazelnut	4	2	B	B	E	E
7	Hazelnut	4	ANA	B	ANA	E	ANA
8	Hazelnut	5	ANA	B	ANA	E	ANA
9	Hazelnut	6	ANA	B	ANA	E	ANA
10	Hazelnut	6	NP	B	NP	E	NP
11	Hazelnut	6	INC	B	INC	E	INC
12	Hazelnut	7	NEG	S	NEG	R	NEG
13	Hazelnut	7	NEG	O	NEG	R	NEG

Note: Comparison of DBPCFCs with positive outcome (active only) and related open food challenges (n = 33). SD is presented as dose step (1–7) with an R added to the number in case the dose step was repeated. In case SD is different between both food challenges of the same patient, this are highlighted with blue colour. If SD difference is more than one dose step, this is highlighted with orange colour. Further details can be found in Table S7.

Abbreviations: ANA, Open food challenge not performed due to anaphylaxis during DBPCFC; B, Both objective and subjective symptoms; DBPCFC, Double‐blind placebo‐controlled food challenge; E, Food should be eliminated from the diet; INC, Open food challenge was performed, and outcome was inconclusive; NEG, Open food challenge was performed, and outcome was negative; NP, Open food challenge not performed at request of the study participant; O, Solitary objective; R, Food can be reintroduced (until pre‐defined maximum amount); S, Solitary subjective; SD, Stop dose.

4. Discussion

This study shows that the diagnostic value of open food challenges, performed following the latest guidelines, is non‐inferior to DBPCFCs for children 4 years and older suspected to be cashew nut, hazelnut, or peanut allergic. Based on our results, we conclude that open food challenges can accurately predict DBPCFC outcomes. Our results support implementing open food challenges in clinical practice for these children, which would shorten time to diagnose these food allergies, reduce healthcare usage and costs, and be less burdensome for children and their carers.

According to recently updated guidelines, DBPCFC is preferred in the following cases: suspected non‐immune‐mediated reactions, expected non‐specific symptoms, and for anxious patients [5]. DBPCFC is also indicated for research purposes, although open food challenges were proposed to be suitable for research with infants and young children, since psychological and subjective issues are thought to be less present or evident [5]. Our findings suggest that previously mentioned restrictions to perform open food challenges are no longer necessary for children with suspected allergies to peanuts, cashew nuts and hazelnuts. If comparable results are found in at least one similar diagnostic trial, (inter‐)national guidelines may be adapted accordingly.

To the best of our knowledge, this study is the first to prospectively compare DBPCFC and open food challenge outcomes, performed following recent guidelines. The strength of our study is that the same child underwent both food challenges for the same allergen using the same recipe within a predefined time frame, which allowed us to compare challenge outcomes, as well as eliciting and stop doses. Unforeseen circumstances caused deviation of the time frame between the challenge days in a few exceptional cases (Table S4). Objective assessment of both OFCs was guaranteed throughout the study, as for each individual patient, both challenges were conducted by a different nurse. Given that OFC outcome relies on interpretation of symptoms, involving at least two different nurses for both challenges ensured independent symptom registration. This approach facilitated an objective comparison of both challenges.

We acknowledge that the occurrence of symptoms during DBPCFC might have affected patients’/caregivers' expectations of the subsequent open food challenge. This is illustrated by the fact that of the scheduled open food challenges, 6/63 were not performed as requested by study participants. We chose not to schedule OFCs in random order, as we anticipated that patients and/or their parents might be unwilling to perform a second OFC if symptoms occurred during the first challenge, as previously mentioned during interviews [10]. Moreover, DBPCFC outcomes were not reported to the patient until after the open food challenge; thus, uncertainty on DBPCFC results may possibly reduce expectation bias for the second OFC. Although the formal outcome was communicated only after both challenges had been performed, patients were aware of whether they experienced any symptoms. This awareness may have reduced the likelihood of false‐positive outcomes in the open food challenge. However, a similar effect could also apply to the DBPCFC outcomes if the open food challenge was conducted first. We speculate that the drop‐out percentage might have been higher if both challenges would have been performed in random order, especially when an open challenge, potentially resulting in allergic symptoms, is performed first. Completion of the placebo day of DBPCFC, during which no symptoms are expected, could result in growing confidence of participants if performed prior to the open challenge. To minimise the effect of experienced symptoms, DBPCFC outcomes were kept blinded for patients, caregivers, and nurses involved with performed food challenges until the specific patient had undergone the third and final challenge day. In two cases, DBPCFC with positive outcomes were followed by a negative open food challenge. Hypothetically, desensitisation could have been induced between both challenges thereby affecting the outcome, but it is presumed that a longer period of exposure is necessary to achieve desensitisation [18]. Both challenges were performed for hazelnut, and both patients were also sensitised to pollen. Results might have been influenced by cross‐reactive allergy with pollen, although all challenge days for each patient were performed within the same season, minimising a potential influence of the season.

Although challenge outcomes were comparable, the percentage of solitary subjective symptoms during open food challenges was higher in comparison to DBPCFCs (46.4% and 12.1%, respectively). This intriguing finding may be explained by the fact that patients were more anxious in case an allergic reaction occurred during the previous DBPCFC for the same food, although this was not formally investigated as part of our study. On the other hand, healthcare professionals might have been more aware during the performance of the open food challenge since everyone involved knew that the suspected food allergen was administered. However, all reported solitary subjective symptoms (i.e., itchy mouth and abdominal pain/nausea) are common among allergic patients and might function as physical warning signals. Recently, it was found that the occurrence of solitary subjective oral symptoms might lead to falsely diagnosing patients to be food allergic in 30% of performed OFCs. If those patients were continuously exposed to the suspected food allergen, they were not at risk for more severe allergic reactions [19]. However, even in the absence of severe and possibly life‐threatening symptoms, patients might feel uncomfortable or even anxious when experiencing these subjective symptoms. Thus, positive OFC outcomes based on solitary subjective symptoms are in line with symptoms experienced by patients in daily life. On the other hand, more severe reactions occurred during DBPCFC. The patients who have never eaten the food to which they are sensitised are not yet aware of how the symptoms will develop during DBPCFC. This could explain why the symptoms during DBPCFC were more severe than during the open food challenge, as they were more alert to the onset of symptoms at that time.

ED was similar between matched DBPCFCs and open food challenges. Specifically, our results showed the same dose for 7/24 challenges and different EDs (i.e., first symptoms did occur at different dose steps) for 14/24 positive open food challenges compared to related DBPCFCs. Hence, results of both OFCs may be comparable, yet based on different symptoms and/or different eliciting (threshold) doses. This might be explained by awareness, as everyone involved knows the suspected food allergen is administered and because of the experience of the previously performed DBPCFC. As a result, children are more aware and better able to recognise possible symptoms. Variation in ED may also be attributed to the well‐known influence of co‐factors (e.g., use of medication, exercise, etc.) on the occurrence of allergic reactions [20, 21]. We believe that the challenge outcome is most relevant in determining related dietary advice. After all, the occurrence of a mild symptom at a higher dose during standardised OFC can, due to various co‐factors, escalate to a moderate or severe symptom at a lower dose in everyday life.

Several limitations warrant consideration. Firstly, our results apply to cashew nuts, hazelnuts and peanuts. Currently, open food challenges are increasingly performed all across Europe, mostly because of practical reasons in comparison to the more labour‐intensive DBPCFC [5]. Given the need to compare the diagnostic accuracy of DBPCFCs and open food challenges, further research similar to our study should be performed for other food allergens and other age groups. Post hoc analysis (Table S6) suggested lower diagnostic performance for hazelnut compared to cashew nut and peanut, which warrants further investigation. We further recommend investigating cow's milk, egg, and wheat, because these are the most common food allergies in children and essential for the child's diet. Secondly, although we did not communicate the formal outcome of DBPCFC to children and parents before finalising the open food challenge, they knew if symptoms occurred or not, and this may have reduced the false‐positive outcome of the open food challenge. This potential bias might be diminished if both challenges are randomised. However, based on interviews with parents of (suspected) allergic patients, it became clear that parents might be hesitant to continue further study participation if symptoms had occurred. Therefore, we decided to start with DBPCFC, which includes at least one day (placebo) on which no symptoms are expected. Thirdly, during the open food challenge, the recipe for the active day of DBPCFC was used to eliminate possible matrix differences. Hypothetically, the outcome could be affected since open food challenges are often performed by administering the suspected food allergen in its natural form, potentially increasing anxiety and expectation bias. Nevertheless, we anticipated that symptoms would occur if an individual was allergic, irrespective of the fact that the allergens were concealed.

In summary, this study shows that the outcome of DBPCFCs and open food challenges is non‐inferior for children suspected to be allergic to cashew nuts, hazelnuts and peanuts. If additional studies support our findings, less demanding open food challenges may be routinely integrated into daily practice. Additionally, more research is needed to investigate similar comparisons for other major food allergens. Ultimately, this may result in a decreased reliance on DBPCFCs and the implementation of open food challenges for diagnosing food allergies for selected allergens in the future.

Author Contributions

W.W.W., A.B.S., C.E.M.H., G.N.M., J.M.V., G.H.K., and A.W.A.K. conceptualised and designed the study. W.W.W. analysed the data. W.W.W., A.B.S., C.E.M.H., G.N.M., J.M.V., G.H.K., and A.W.A.K. contributed to data interpretation. W.W.W. wrote the first draft of the manuscript. W.W.W., A.B.S., C.E.M.H., G.N.M., J.M.V., G.H.K., and A.W.A.K. were involved in critically reviewing and editing the paper. A.W.A.K. supervised the study.

Conflicts of Interest

A.B.S. reports research grants from Nestle Research, Lausanne, and Aimmune, outside the submitted work. Her institution received compensation for her consultancy from Sanofi Netherlands and Nestle Research, Lausanne. G.H.K. reports grants from the Netherlands Lung Foundation, ZON‐MW (VICI grant), European Union (H2020), Vertex, GSK, and TEVA in the Netherlands, outside the submitted work. His institution received compensation for participation in advisory board meetings for AZ and PURE IMS and for invited lectures from Sanofi, Boehringer Ingelheim, and AZ. G.H.K. is the founder and chair of the exquAIro Foundation, outside the submitted work. The other authors report no proprietary or commercial interest in any product mentioned, concept discussed, or personal relationships with other people or organisations that could influence their work and conclusions in this article.

Supporting information

Data S1.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.