Food-Related Symptoms and Food Allergy in Swedish Children from Early Life to Adolescence

Background Risk factors for persistence of food-related symptoms (FRS) and food allergy (FA) from early life to adolescence are incompletely understood. The aim of this study was to identify risk factors for FRS and FA in adolescence amongst children with FRS or FA in the first four years of life (early life). Methods In children enrolled in a Swedish birth cohort and followed to 16 years (n = 2572), we defined children with early life FRS in the absence of FA, and FA. Corresponding phenotypes were defined at 16 years. Associations between potential risk factors at 4 years and FRS and FA at 16 years were investigated using logistic regression. Results Early life FRS and FA prevalences were 12.2% and 6.8%, respectively. Amongst children with early life FRS, 35.7% had FRS or FA at 16 years, whereas 74.3% of the children with early life FA had FA at 16 years. For each of the early life phenotypes, parental allergy, early life allergic multimorbidity, early life reactions to peanuts/tree nuts and IgE reactivity at 4 years were statistically significantly associated with FRS or FA at 16 years. In contrast, male sex was associated with an increased risk of FA at 16 years among children with early life FA only. Conclusions In early life, food-related symptoms are twice as common as food allergy. Unlike food allergy, food-related symptoms often remit by adolescence. Yet, these phenotypes have many common risk factors for persistence to adolescence.

Risk factors for early life FRS and FA have been studied. Family history of allergy [23][24][25] and allergic diseases in early life, particularly eczema [23] and Immunoglobulin E (IgE) reactivity [24][25][26][27], are established risk factors, whereas early life environmental factors and sociodemographic exposures remain incompletely understood [23,[25][26][27][28][29][30]. Less is known about the risk factors for, and the prognosis of FRS and FA from early life through adolescence. Therefore, we aimed to identify risk factors for FRS and FA in adolescence amongst children with FRS or FA in the first four years of life (early life).

Study design and population
This study is based on data from the BAMSE project [30], a longitudinal, population-based birth cohort of 4089 children born in Stockholm, Sweden between 1994 and 1996. Parents completed questionnaires at baseline (children 2-3 months old), and 1, 2, 4, 8, 12 and 16 years old. The response rate through 16 years was 78% (3181/4089) from baseline.
Children with information on parent-reported FRS and doctor-diagnosed FA at 1, 2, 4 years, and 16 years were included in the present study (n = 2572, 62.9% of the entire cohort). Information on allergen-specific IgE reactivity in serum was available in a subpopulation of participants at 4 years and at 16 years (n = 1903 and n = 2057, respectively), including 1617 participants for whom IgE information was available at both 4 and 16 years.

Definitions of FRS, FA, systemic reactions, anaphylaxis and markers of severity
Food-related symptoms (FRS). Early life FRS: Children were classified as having early life FRS if their parent(s) reported a specific symptom(s) to a specific food(s) in the 12 months prior to the 1, 2 and/or 4-year questionnaires [31], but not doctor-diagnosed FA at any of these time points. Specific symptoms included vomiting, diarrhoea, eczema, urticaria, itch or swelling of the lips and/or eyelids, runny nose, and/or asthma. Specific foods included milk, egg, fish, tree nuts, peanuts, peas, soy, wheat, fruit with pits/pips, and at 4 years only, citrus, chocolate and banana. Foods reported to cause a reaction were labelled culprit foods. FRS at 16 years: Similar to the definition in early life, this variable was based on parentreported specific symptom to a specific food in the 12 months prior to the 16 year assessment, without doctor-diagnosed FA. Children were classified as having FRS if they continued to avoid a specific food due to a previous reaction(s) and/or due to results of previous allergy testing and did not have a doctor's diagnosis of FA.
Food allergy (FA Anaphylaxis at 16 years. This variable is based on parent-reported data at 16 years corresponding to the currently accepted definition of anaphylaxis [32]; see S1 Table. Briefly, anaphylaxis was considered as any reaction involving at least two organ systems (gastrointestinal, dermatological, lower respiratory, cardiovascular) with specific, predefined symptoms. This variable was defined for both FRS and FA at 16 years [32].
Markers of severity of FRS and FA. Age at first reaction: Amongst those with early life FRS or FA, this variable was defined as first reaction by 2 years, or between 2-4 years.
Symptoms: Parent-reported symptoms upon ingestion of food and/or drink, were obtained from the 4 year questionnaire. (Data on symptoms were not collected as the 1-or 2 year questionnaires.) Symptoms were collapsed into three categories: gastrointestinal (vomiting and/or diarrhoea), dermatologic (urticaria and/or oedema) and respiratory (rhinitis and/or wheezing). Children were classified into the most severe category of symptoms reported by their parents.
Specific foods: Parent-reported reactions to individual foods were obtained from the 1, 2 and 4 year questionnaires. Presented herein are the seven most common culprit foods (milk, egg, fish, peanuts/tree nuts, fruits with pips or pits, citrus fruit and chocolate) in our study population. These categories of foods were non-mutually exclusive.
Definitions of baseline characteristics, infant feeding, early life allergic multimorbidity and IgE reactivity Definitions of baseline characteristics, infant feeding, early life allergy-related diseases (asthma, eczema and/or rhinitis) and IgE reactivity are shown in S2 Table. IgE reactivity at 4 years and 16 years. Sera obtained at 4 years and at 16 years were analysed for IgE reactivity to common aeroallergens with Phadiatop1 (cat, dog, horse, house dust mite, timothy, birth, mugwort, mould) and food allergens with fx51 (egg, milk, codfish, wheat, peanut, soy) (Thermo Fisher/Phadia AB, Uppsala, Sweden). Children were considered to have IgE reactivity if they had an IgE antibody level !0. 35

Statistical Analyses
Descriptive statistics included sample sizes, percentages, and 95 th per cent confidence intervals (95%CI). Proportional Venn diagrams were generated to illustrate the progression of FRS, from early life to 16 years.
Associations between exposures (baseline characteristics, infant feeding, early life allergic multimorbidity, markers of severity, IgE reactivity) and outcomes (FRS or FA at 16 years, FA at 16 years) were estimated using binary and multinomial logistic regression, and for which we reported the odds ratios (OR) and 95%CI.
The final models were adjusted for covariates selected based on a priori knowledge and included sex, parental allergy and socio-economic status. In a sensitivity analysis, we additionally included the variable 'early life allergic multimorbidity'. This additional adjustment insubstantially altered (<10%) most point estimates, and did not statistically significantly alter any of the point estimates. As such, the results are not presented herein.

Results
Our study population (n = 2572) was comparable to the entire cohort with respect to participant characteristics, with the exception of small, but statistically significantly higher rates of white collar families (84.7% vs. 82.7%, respectively) and lower rates of tobacco smoking in pregnancy (11.4% vs. 12.9%, respectively) (S3 Table).
Children with early life FRS or FA were comparable to children without reactions or allergy to food regarding distribution of sex, parental migration, socio-economic status or infant feeding (Table 1). Early life asthma, eczema and rhinitis were all more common amongst children with early life FRS (63.4% had at least one of the diseases) and most common amongst children    To further disentangle the association between IgE reactivity at 4 years and FRS or FA at 16 years, we evaluated IgE by aeroallergens and/or food allergens. Among children with early life

Discussion
In this large, population-based study of Swedish children followed from birth to 16 years, early life FRS was approximately twice as common as early life FA. The majority of children with early life FA reported FA also at 16 years, whereas most children with early life FRS did not report symptoms at 16 years. Anaphylaxis at 16 years was more common amongst children with early life FA, compared to children with early life FRS. Parental allergy, early life allergic multimorbidity and early life reactions to peanuts or tree nuts and IgE reactivity were common risk factors for FRS or FA at 16 years amongst children with these phenotypes in early life. Amongst children with early life FA only, male sex was associated with an increased risk. We highlight the strengths of this study. BAMSE is a large, unselected cohort with high retention and detailed follow-up, including types and reactions to foods, through 16 years. To our knowledge, this is the first report on the FRS and FA from early life to adolescence using questionnaire data in combination with sera analyses for IgE reactivity to both aeroallergens and food allergens.
We also acknowledge the limitations of this study. FRS and FA were defined based on parental reports. Further, FA was based on reports of physician diagnosed food allergy. We did not query the type of physician, such as general physician, paediatric allergologist or other, or the type of diagnostic testing or markers of severity, such as levels of mast cell mediators. Thus, we cannot dismiss the possibility of over-reported FA. However, the rates of both early life FRS (12.2%) and FA (6.8%) align with the current global paediatric estimates of 11-22% [2][3][4] and 4-8% [5][6][7] for the two phenotypes, respectively. We also lacked data on the frequency of reactions and treatments in early life, and thus cannot state if either has any predictable influence on FRS or FA in adolescence. Other Swedish studies would support that previously known food allergy and previous prescriptions for adrenaline autoinjectors are a risk factor for emergency department revisits for children with a prior emergency department visit resulting from food reactions [33], but that nearly all (89%) of children are hospitalised only once for allergic reactions [20]. As well, we cannot exclude the possibility that FRS and FA at 16 years were more likely to be reported among children with FRS or FA in early life. Yet, such misclassification is not likely to fully explain the higher prevalence of FRS or FA at 16 years among children with early life FHS or FA, particularly because of the 12-year time lapse between assessments. Further, such misclassification would not explain the observed association between early life risk factors and FA or FRS at 16 years. Lastly, we acknowledge that many of the analyses on specific symptoms and foods are based on small numbers, and thus should be interpreted with caution.
Parental allergy was significantly more common amongst children with early life FA compared to early life FRS. This raises the possibility that children whose parents are allergic may be more likely to receive a food allergy diagnosis. However, parental allergy was associated with a two-fold increased risk of FRS and/or FA at age 16 years among both early life phenotypes.
Our finding that males with early life FA had an increased risk of FA at 16 years, compared to females extends the findings from other birth cohorts with follow-up through school-age [26], and parallels findings in which other allergic diseases were considered [34,35]. Yet, females from adolescence onward are much more likely to experience severe reactions, including anaphylaxis, compared to similar-aged men, an observation which has been attributed to estrogen playing a role in the susceptibility to anaphylaxis [36]. Early life allergic mulitmorbidity was a risk factor for FRS and FA at 16 years, possibly with the exception of asthma among children with early life FRS. Notably, the point estimates for early life asthma were substantially different for FRS at 16 years vs. FA at 16 years.
FRS and FA symptoms frequently involve the gastrointestinal and dermatological systems [8,37]. In our study, compared to gastrointestinal symptoms, all other symptoms increased the risk of FA at 16 years, while only the most severe symptoms (i.e. respiratory) were associated with an increased risk of FRS at 16 years. Respiratory symptoms are often [10] but not always [38] associated with severe reactions, including anaphylaxis. In our study, anaphylaxis at 16 years was more common in children with FA in both early life and at 16 years, compared to children with early life FRS and FRS or FA at 16 years (12.3% vs. 1.8%), thereby reinforcing that FA is the more severe phenotype.
Like our study, many [39][40][41][42][43] but not all [15,38,44,45] previous studies on FRS or FA were prospective in design. But, only one previous study included follow-up to adolescence [43]. Most focused on allergies to individual foods [39][40][41][42][43][44][45], particularly milk [39][40][41]45], egg [42,44] and peanut [38,43]. Although these foods warrant attention given the commonality of allergy to these foods amongst children [8], restricting analyses to individual foods may not capture a complete picture of the natural history of FRS or FA from early life through adolescence. In our study of seven foods, we found that children with early life reactions to peanuts/ tree nuts were more than three times more likely to experience FRS or FA at 16 years than children who reacted to milk in early life. Although we cannot conclude that those with FRS or FA at 16 are still allergic to peanuts/tree nuts, our findings align with previous reports that children typically do not outgrow reactions to peanuts [43]. Interestingly, amongst children with early life FRS only, early life reactions to fruits with pips or pits were also associated with an increased risk of FRS or FA at 16 years. This finding extends previous results from our cohort, in which symptoms to pollen and/or fruit by 2 years of age were associated with an increased risk of allergic diseases, including asthma, rhinitis and eczema [46]. Our observation that children with early life FRS are commonly sensitised to aeroallergens only aligns with findings from other studies [3,4] and supports the theory that reactions to fruits with pips or pits are often the result of these cross-reactions. In contrast, IgE reactivity to both aeroallergens and food allergens predominated amongst children with either early life FRS or FA. Peanuts and tree nuts are common triggers of anaphylaxis [8], whereas reactions to fruits with pips and pits tend to involve the oral cavity as a result of cross-reactivity to aeroallergens, such as birch [47]. Yet, systemic reactions and anaphylaxis predominated amongst children with early life FA compared to early life FRS, suggesting that early life reactions ought to be considered in combination with a clinical definition when predicting the severity of future reactions.
Despite several common risk factors for persistence into adolescence, early life FRS and FA appear to be different phenotypes with different prognoses though adolescence. Although our findings must be confirmed in other populations, they may also begin to help answer the question that parents frequently ask: Will s/he outgrow it?
In conclusion, food-related symptoms are twice as common as food allergy in early life. Unlike food allergy, food-related symptoms often remit by adolescence. Yet, these phenotypes have many common risk factors for persistence to adolescence.
Supporting Information S1 Table. Definition of food-induced anaphylaxis at 16 years, based on criteria from the National Institute of Allergy and Infectious Disease/Food Allergy and Anaphylaxis Network. Ã Based on Reference 32. †Excludes information on blood pressure, heart rate or oxygen saturation, as these data were not collected in the BAMSE study. (DOCX) S2 Table. Definitions of baseline characteristics, infant feeding, early life allergy multimorbidity and Immunoglobulin E (IgE) activity. (DOCX) S3 Table. Distribution of baseline characteristics and infant feeding of the entire cohort and the study population. Ã Parent.report of doctor-diagnosis of asthma and/or hayfever in combination with allergy to furred pets by either or both parents at time of enrolment. †Mother reporting smoking 1+ cigarettes per day during pregnancy. Bold text denotes statistically significant results. (DOCX) S4 Table. Children with early life food-related symptoms (FRS) or food allergy (FA) and for whom IgE-reactivity data at 4 years were available in relation to food IgE reactivity-associated FRS or FA to 16 years. (DOCX)