Food allergy outside the eight big foods in Europe: A systematic review and meta‐analysis

Abstract Background The 2014 estimates of prevalence of food allergy (FA) in Europe published by the European Academy of Allergy and Clinical Immunology included only the eight so‐called big foods (cow's milk/egg/wheat/soy/peanut/tree nuts/fish/shellfish). Those estimates have recently been updated. Complementing this, we sought to identify and estimate the prevalence of allergy to other foods that have been reported during the last decade. Methods Six databases were searched for studies published 2012–2021. Random‐effects meta‐analysis was performed to derive pooled prevalence of allergy to each food. Results Twenty‐seven studies were included, containing a total of 66 FAs. Among the most frequently reported FAs, the lifetime and point prevalence range of self‐reported kiwi allergy was 0.1%–1.0% and 0.2%–8.1%, respectively, while the food challenge (FC)‐verified kiwi allergy point prevalence range was 0.01%–0.10%. The point prevalence range for self‐reported peach allergy was 0.2%–3.2%, while the range for FC‐verified peach allergy was 0.02%–0.05%. The lifetime and point prevalence range for self‐reported tomato allergy was 0.01%–1.8% and 0.2%–2.1%, respectively. Conclusion Allergy to some foods traditionally not considered important are now emerging as relevant FAs. The focus on FA in Europe should not be limited to the so‐called eight big FA, but extended to other types of foods which need to be considered both for clinical purposes and population risk assessment.


| INTRODUCTION
Food allergies (FAs) have become a common topic for healthcare systems, as the incidence and prevalence have reportedly increased over the last decades.However, there is a need to improve this evidence base in order to gain a better appreciation of the frequency of FA across Europe, through which its healthcare and societal burden can be elucidated more clearly.A decade ago, the European Academy of Allergy and Clinical Immunology (EAACI) published a systematic review and meta-analysis on the prevalence of food allergy/sensitization (FA/FS) in Europe based on studies published between 2000 and 2012, but the focus was on allergy to the eight so-called big foods (i.e., cow's milk, hen's egg, wheat, soy, peanut, tree nuts, fish, and shellfish). 1,24][5] Indeed, many other foods can elicit FA/FS, including fruit, vegetables/legumes, cereal, meat, and others.
An additional objective of our update was to identify and estimate the prevalence of FA/FS in foods other than the eight big foods that have been reported during the last decade.The current work therefore summarizes the available evidence on FA/FS to foods other than the eight so-called big foods in Europe and estimates their prevalence, where data allow.

| Protocol registration, search strategies, and study identification and selection
Before starting the systematic review, a protocol was registered on the International Prospective Register of Systematic Reviews (PROSPERO) (registration number CRD42021266657).The search strategy was adapted from the methodology employed in the previous EAACI systematic review and meta-analysis. 1,2In brief, we combined the two concepts of FA/FS and epidemiology to detect all relevant literature.We searched six electronic databases (MEDLINE, Embase, CINAHL, Web of Science, Cochrane Library, and Scopus) to collect papers and conference abstracts on FA/FS in Europe published between September 2012 and June 2021.We retained all keywords employed in the 2014 reviews; additional keywords were included to avoid missing any relevant studies, as well as to account for developments that have occurred in the chosen databases in the last 10 years.The full search strategy has been reported in the Supporting Information S1 of our previously published paper on the frequency of any FA in Europe. 3We did not apply any language restriction to the database search.Studies not written in English were translated by a member of our research team fluent in the language.In case translation was not possible, but an English abstract was available, relevant data were extracted from the paper abstract, while at the same time employing Google Translate to translate the text.Having a clear idea of the paper content from the abstract summary allowed us to limit the risk of data misinterpretation.
All studies that investigated FA/FS in the general European population, of any age and gender, were considered eligible.
The following study types were eligible for inclusion: systematic reviews, cross-sectional studies, cohort studies, case-control studies, clinical trials and routine healthcare studies.Narrative review, discussion papers, non-research letters or editorials, case-series, casestudies, and animal studies were excluded.Relevant papers were screened by four independent reviewers, working in pairs (SN/GS and YA/MA), first by title and/or abstract, and later by full text.In case of disagreement between the pairs, conflicts were resolved with consensus or after consulting the project PI (BN).We employed the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) flow diagram to describe the screening process and we conducted our research according to the guidelines of the PRISMA 2020 Statement. 6

| Outcomes
The systematic review aimed to provide up-to-date estimates on the incidence, prevalence, and time trends for FAs outside the eight big FAs in Europe for the period 2012-2021.An allergy to any food outside the eight big foods were considered in the analysis, including fruits, vegetables, legumes, meat, cereals (not including wheat), seeds, herbs, spices, and condiments, and many other food types.[3][4][5] We could not differentiate between immunoglobulin E (IgE)mediated and non-IgE-mediated FA, as this was not usually differentiated in the included studies.
The following outcomes were included: (1) Lifetime prevalence

| Risk of bias assessment
The risk of bias assessment for individual studies was carried out by the same pairs of reviewers who completed the screening procedure by employing an adapted version of the Critical Appraisal Skills Programme (CASP; http://www.casp-uk.net)quality assessment tool.
Conflicts between the pairs were resolved by consensus or by consulting the project PI (BN).

| Data analysis and synthesis
Data from each included study were collected using a customized data extraction form.When sufficient data were available, we recalculated the estimates using minimally measured events rather than extrapolated ones.Meta-analysis was considered meaningful for all outcomes with three or more records available.To obtain the 95% confidence intervals (95% CI), we employed the Wilson score method without continuity correction. 7In case of the need of clarification regarding the data presented in a study, a request of clarification was sent to the corresponding author of the said paper.Following the revised protocol of the 2014 review, we attempted to stratify the available data by age groups (children [0-17 years], and adults [≥18 years]), and by European regions (Northern, Eastern, Southern, Western Europe) according to the classification by the United Nations (see Appendix 1).For this review, we analyzed the geographical distribution of the studies reporting on FA/FS only for foods that were reported in at least four studies (equaling the total number of European regions according to the United Nations classification, i.e., four regions).
We performed random-effects meta-analysis to derive pooled prevalence estimates for individual FAs from all studies that provided adequate numerical data and shared methodologically comparable data during 2012-2021.Because of the scarcity of data, it was not possible to perform meta-analysis by age and by European region.
We employed the Stata software (StataCorp.2019, Stata Statistical Software: Release 16.StataCorp LLC) to complete the analysis and used the I 2 statistic to assess heterogeneity across studies.We included European countries within and without the Organization for Economic Cooperation and Development (OECD) in the metaanalysis.
We further compared the pooled prevalence estimates obtained from random effect meta-analysis for FA/FS to foods other than the eight big foods with our recently published 2012-2021 pooled prevalence estimates for FA/FS to the eight big foods. 4

| RESULTS
The study selection and screening process is presented in the PRISMA flow chart (Figure 1).We identified 38,903 records published between 2012 and 2021.After the screening process, 27 studies reporting on FA/FS other than the eight big FAs were included.  Tab 1 summarizes the main characteristics (i.e., age of the subjects involved, type of study, etc.) and the overall risk of bias score of the included studies.

| Frequency of FA
The detailed results on the prevalence and incidence of the FAs are summarized in Tables S1 and S2.
Table 2 summarizes the ranges of prevalence and incidence for all the FAs identified in this review using different assessment methods (i.e., point prevalence self-reported FA, point prevalence sIgE sensitization).For synthesis purposes, we have divided the different foods into the following five categories: fruits, vegetables/ SPOLIDORO ET AL. legumes, herbs/seeds/condiments/spices, cereals, meat, and other foods.
The total number of reported FAs outside the eight big FAs was 66. Kiwi was the most studied food (reported in 12 studies), followed by peach and tomatoes (9 studies each), apple, banana, and sesame seed (7), strawberry and chocolate (6), carrots (5), celery, lentils, and beef (4).Thirty-seven of the 66 foods were reported in only one study, among them being various types of meat (e.g., red meat, pork, salami), specific types of fruits (e.g., nectarine, plum), vegetables (e.g., broccoli), as well as different types of condiments (e.g., chili).Twelve foods were reported in at least four studies, that is, kiwi, peach, tomato, sesame seed, apple, banana, strawberry, chocolate, carrot, celery, lentils, and beef.Figure S2 presents the distribution by the European region of the studies reporting on these 12 FA/FS.Meta-analysis was performed for 19 foods for which enough data were available to allow pooling of results for at least one of the outcomes investigated (i.e., point prevalence self-reported FA, FCverified FA etc.).The pooled lifetime and point prevalence estimates for each FA/FS according to the different outcomes investigated are presented in Figures 2-8.There was significant heterogeneity among the studies pooled in the meta-analysis (I 2 ≥ 80 in each case).Due to the paucity of data, we could not observe any consistent pattern across age groups.

| SPT or sIgE sensitization
Ten studies reported FA by means of sensitization (positive SPT or sIgE test) to foods, and two were international multi-center studies.
Six studies reported estimates for sIgE positivity, while seven reported estimates for SPT positivity.For positivity in sIgE tests, the FA with the highest reported point prevalence estimates was banana (15.4%), while the lowest prevalent were kiwi, black pepper, strawberry, and honey (each at 0.01%).

Type of reported food allergy/sensitization
Overall risk of bias assessment
Data on SPT positivity were too scarce to perform a metaanalysis.

| Challenge-verified food allergy
From the 10 studies that performed food challenges (OFC or DBPCFC), the highest prevalence estimate was for beef allergy (0.3%), while the lowest was negligibly for blueberry allergy.
Data on FC sensitization were too scarce to perform metaanalysis.

| Comparison of pooled prevalence estimates for food allergy or sensitization outside the eight big foods and pooled estimates for the eight big food allergies
Box S1 in the Supporting Information S1 presents the comparison between the pooled prevalence estimates obtained for FA/FS outside the eight big foods and our recently published pooled prevalence estimates for the eight big FA/FS. 4We have compared estimates for lifetime prevalence of self-reported FA and of self-reported physician-diagnosed FA, as well as for point prevalence of self-reported FA and of sIgE-positive FA/FS.Lifetime prevalence of selfreported physician-diagnosed FA was higher for the commonly studied cow's milk, egg, fish, peanut, tree nuts, wheat, soy, and shellfish FA/FS (i.e., the eight big FA/FS) compared to other foods.For all the other outcomes (i.e., lifetime prevalence of self-reported FA, point prevalence of self-reported FA and of sIgE positive FA/FS), we did not observe a clear pattern in terms of highest/lowest prevalence between FA/FS and the eight big foods and FA/FS to other foods.

| Statement of principal findings
For the first time, this systematic review provides a comprehensive view on the frequency of FA/FS outside the eight so-called big FAs (i.e., cow's milk, egg, wheat, soy, peanut, tree nuts, fish, and shellfish allergy).The most frequently investigated food was kiwi.The overall pooled estimates for kiwi allergy or sensitization was 0.4%, 3.4%, and 5.8% for self-reported lifetime, self-reported point, and point prevalence of sIgE sensitization, respectively.Among vegetables, carrot allergy or sensitization was the most reported, with pooled prevalence of 1.5% and 6.6% for point prevalence self-reported allergy and sIgE sensitization, respectively.Beyond fruits and vegetables, other frequently investigated foods were sesame seeds (7 studies), chocolate (6), lentils (4), and beef (4).Sesame seed, which is included F I G U R E 6 Point prevalence of immunoglobulin E sensitization to fruits in Europe.

| Strengths and limitations of the current update
This systematic review and meta-analysis was conducted following a rigorous methodology and a systematic approach in every phase of the study.We completed a comprehensive literature search of six major electronic databases, including two databases which were not previously considered in the 2014 review.Moreover, compared to the 2014 review, we included more keywords in the database search to limit the possibility of missing any relevant information.Including We included all methods of assessment of FA or FS and did not apply any language restriction during database searches and literature screening, which allowed us to provide a comprehensive picture F I G U R E 7 Point prevalence of immunoglobulin E sensitization to grains, legumes, and vegetables in Europe.
-17 of 22 of the frequency of allergies to foods beyond the eight big FAs in Europe.
Conversely, the quality of our study is partly constrained by the limited number of data available for some of the allergenic foods investigated, including some foods that are commonly perceived as important allergenic foods.For instance, although lupins are listed among the 14 food ingredients that need to be labeled under the EU 1169/2011 regulation, only one study reported on the prevalence of F I G U R E 8 Point prevalence of immunoglobulin E sensitization to seeds in Europe.
lupin allergy/sensitization by measuring sIgE positivity. 20Similarly, mustard seed allergy/sensitization, which is also included among the food ingredients regulated by the EU 1169/2011, was investigated only by means of sIgE positivity and by sIgE positivity plus allergic symptoms in two studies.]26 As already observed in the systematic review published on the prevalence of any FA/FA to the eight big foods, the high prevalence observed for self-reported FA and for sIgE/SPT FS for some of the foods investigated did not match the low prevalence estimates obtained by measuring FC-verified allergy.However, even though FC is considered the gold standard methodology to assess FA, its use was very restricted in the studies included in the review.The correct assessment of FA would be improved through increased use of the gold standard FC measurement.
The high heterogeneity observed across studies may be a sign of methodological discrepancy in the way FA/FS was addressed by the authors of the included studies (e.g., different methods of assessment for the considered outcomes), but it could also indicate that there is a wide variation in the prevalence of FA/FS within and between ages and European regions.Considering these two possible interpretations for the encountered high heterogeneity, the data resulting from the analysis of the studies should be evaluated with caution.
One of the limitations of the review is that most of the selected studies did not differentiate between IgE and non-IgE allergy, which prevented us from distinguishing between IgE-mediated and non-IgE-mediated FA phenotypes.
Finally, albeit illogical, pooled estimates for lifetime prevalence of self-reported FA to apple, kiwi, strawberries, and tomato were lower than point prevalence.This inconsistency can however be explained by the fact that for most of the foods investigated, lifetime and point prevalence were not pooled from the same studies.

| Comparison of findings to previous studies on allergy or sensitization to foods other than the eight big foods
To our knowledge, this is the first systematic summary of the prevalence of allergy or sensitization to foods other than the eight so-

| Comparison of findings to previous findings on allergy or sensitization to the eight big foods
Although cow's milk, egg, wheat, soy, peanut, tree nuts, fish, and shellfish are traditionally considered the eight most common allergenic foods in Europe, pooled prevalence estimates for the eight big FA/FS were not always the highest compared to FA/FS to other foods in Europe between 2012 and 2021.Indeed, especially for point prevalence of self-reported FA, and for point prevalence of sIgEpositive FA/FS, some of the foods investigated in this review were found to have a higher prevalence than many of the eight big foods.
However, the number of available prevalence estimates for foods other than the eight big foods was frequently lower than the available estimates for allergy or sensitization to the eight big foods, which can result in a less accurate meta-analysis.

| Interpretation and implication of the current findings
As already mentioned, the results of the systematic review and metaanalysis should be interpreted with caution.The fact that most of the studies were graded as at moderate risk of bias according to the CASP quality assessment tool, combined with the high heterogeneity observed across studies, prevents the derive of univocal conclusions on the study findings.
More studies would be required to obtain a comprehensive view of the burden of FA outside the eight big FAs.Moreover, similar to what was observed in the updated review of any FA and common FA, future studies would benefit from the definition of shared protocols and standardized methodologies for the assessment of FA in Europe.

| CONCLUSIONS
The data presented in this review suggest that allergy to some foods traditionally not considered important are now emerging as relevant FAs, including but not limited to kiwi, peach, tomato, sesame seed, apple, banana, strawberry, chocolate, carrot, celery, lentils, and meat.
This observation may partly explain our recent finding that, although the prevalence of "any FA/FS" has increased in the last 10 years, the prevalence of FA/FS in cow's milk, egg, wheat, soy, peanut, tree nuts, (i.e., prevalence of subject reporting ever having a reaction or hypersensitivity to respective foods) and point prevalence (i.e., prevalence of subjects reporting having a reaction or hypersensitivity to respective foods currently or during the past 12 months) of selfreported FA; (2) lifetime and point prevalence of self-reported physician-diagnosed FA (i.e., doctor-diagnosed FA reported by a subject in a questionnaire); (3) point prevalence of specific immunoglobulin E (sIgE) sensitization; (4) point prevalence of skin prick test (SPT) sensitization; (5) point prevalence of symptoms plus sIgE sensitization; (6) point prevalence of symptoms plus SPT sensitization; (7) point prevalence of food challenge (oral food challenge [OFC] or double-blind placebo-controlled food challenge [DBPCFC]) positivity; and (8) point prevalence of food challenge positivity (OFC or DBPCFC) and/or clinical history of FA (i.e., FA confirmed by a convincing clinical judgment by a physician without food challenge).

F I G U R E 1
PRISMA flow diagram for systematic review on prevalence of food allergy outside the eight big FAs between September 2012 and June 2021.T A B L E 1 Main characteristics and overall risk of bias of the studies investigating the prevalence and incidence of FAs outside the eight big FAs in Europe between September 2012 and June 2021.

T A B L E 2
Abbreviations: DBPCFC, double-blind, placebo-controlled food challenge; FC, food challenge; OFC, oral/open food challenge; PTP, prick by prick skin test for sensitization to specific foods; sIgE, specific immunoglobulin E test; SPT, skin prick test for sensitization to specific foods; SR, self-reported FA.a Cumulative incide.

16 of 22 -
SPOLIDORO ET AL.new databases and keywords allowed us to take into consideration the advances that have been made since the publication of the 2014 systematic review.

country Study design Study population Age of subjects Method of outcome assessment Type of reported food allergy/sensitization Overall risk of bias assessment N (children/adults) number included
-SPOLIDORO ET AL.T A B L E 1 (Continued)Reference,

of 22 3.2 | Self-reported food allergy
In total, 19 studies reported on self-reported FA, one of which was an international multi-center study reporting multiple estimates for each FA/outcomes investigated (one from each center/country).The highest prevalence of self-reported FA was 18.1% for the lifetime prevalence of citrus fruit and 10.1% for the point prevalence of Point prevalence of self-reported allergy to fruits in Europe.Point prevalence of self-reported allergy to carrots, chocolate, and vegetables in Europe.
CI 4.19, 12.03), apple (7.03%,95%CI4.70,9.35),tomato(6.32%, 95% CI 4.12, 8.52), kiwi (5.80%, 95% CI 3.16, 8.44), and melon (4.32%, 95% CI 2.49, 6.16) (Figure6).Pooled point prevalence of sIgE sensitization was only available for carrot (6.63%, 95% CI 4.57, 8.69), celery (6.67%, 95% CI 4.54, 8.80), and lentils (4.68%, 95% CI 3.18, 6.18) within the vegetable/legumes category, and for corn (5.42%, 95% CI 3.54, 7.30) within the cereal category (Figure7).Finally pooled point prevalence sIgE sensitization was also available for F I G U R E 3 SPOLIDORO ET AL.F I G U R E 4 Lifetime prevalence of self-reported physician-diagnosed allergy to fruits in Europe.among the foods that need to be labeled on pre-packed foods when used as ingredients under the EU 1169/2011 regulation, gave a pooled prevalence of 6.1% for sIgE sensitization.Most other FAs/FS were very rarely reported, 37 only in one study, which limited our ability to derive a clear picture of their prevalence in Europe.Given the paucity of data, no consistent patterns in prevalence could be seen by European region and between children and adults across the F I G U R E 5Finally, most studies were rated as having a "moderate" risk of bias, indicating that the methodology through which evidence-based data on FAs are derived needs to be improved, in order to gain a better appreciation of the frequency of FA across Europe.SPOLIDORO ET AL.
38lled big foods.Before our study,Zuidmeer et al. (2008)published a systematic review of the worldwide prevalence of plant FA, defining six plant food categories, that is, fruits, vegetables/legumes, tree nuts, soy, wheat, and sesame seed/cereals/spices/herbs.38For all the other food types within the categories investigated by the authors, only one estimate was available.Notably, although the authors investigated plant FA worldwide, most of the estimates available came from European studies.Another interesting observation is that, although Zuidmeer et al. investigated four FA outcome, that is perceived FA, sIgE-positive FS, SPT-positive FS, and FC-verified FA, for the fruit, vegetables/legumes and sesame seed/cereals/spices/herb category, there were no available data on sIgE sensitization, while more data were available on SPT sensitization.This result is almost opposed to what was observed in our review where sIgE sensitization was far more represented than SPT, which may suggest that in vitro food specific IgE testing is now preferred over SPT, although the two methodologies have different sensitivity and specificity and do not always concord.