Combinations of self‐reported rhinitis, conjunctivitis, and asthma predicts IgE sensitization in more than 25,000 Danes

Abstract Background Allergic rhinitis (AR), allergic conjunctivitis (AC), and asthma composing multiple phenotypes and improved understanding of these phenotypes and their respective risk factors are needed. Objectives The objective of this study was to define the prevalence of AR, AC, and asthma and their association with allergen‐specific immunoglobulin E (sIgE) sensitization in a large cohort of blood donors and identify risk factors. Methods From the nationwide population‐based Danish Blood Donor Study, 52,976 participants completed an electronic questionnaire including AR, AC, asthma, allergic predisposition, and childhood residence. Of these, 25,257 were additionally tested for sIgE to inhalation allergens (Phadiatop). Results The prevalence of sIgE sensitization, AR, AC, and asthma was 30%, 19%, 15%, and 9%, respectively. The youngest birth cohorts had the highest prevalence of sIgE sensitization and symptoms of asthma, AR, and AC, and for asthma, they apparently experienced symptoms at an earlier age. The sIgE sensitization was positively associated with male sex. The sIgE seroprevalence was higher in participants with both AR and AC (ARC) than in participants with either AR or AC. Allergic predisposition and sIgE sensitization increased the risk of the diseases, while farm upbringing was associated with reduced prevalence of ARC, however, only in sIgE sensitized participants. Conclusion Birth year, childhood residence, sIgE sensitization, and allergic predisposition were associated with asthma, AR, and AC prevalence. Individuals with self‐reported ARC represent a primarily sIgE‐positive phenotype, while those with either AR or AC represent more diverse phenotypes.

blood. 26 Questions regarding allergy, asthma, and place of upbringing were based on standardized questions from the European Community Respiratory Health Survey (ECRHS) questionnaire. 27 AR was defined as an affirmative answer to the question: "Do you have any nasal allergies including hay fever?", AC as an affirmative answer to the question: "Do you have any eye allergies including hay fever?", and ARC as both AR and AC. Asthma was defined as an affirmative answer to the question: "Do you have or have you ever had asthma?".
Allergic and asthmatic participants stated their age at the time of the first symptoms (in 5-10-year age intervals) and months with symptoms.
Based on questionnaire data, the study population was divided into distinct single disease and multimorbidity groups (the coexisting of asthma, AR, and/or AC). However, "only AR" and "only AC" were merged for most analyses to avoid low numbers and insufficient power in these groups. This resulted in six distinct phenotype groups: asthma without AR and without AC; asthma with either AR or AC; asthma with ARC; ARC without asthma; either AR or AC without asthma; and controls (no AR, no AC, and no asthma). For distribution of phenotypes, see Figure A1.
Body mass index (BMI) (kg/m 2 ) was calculated from self-reported weight and height, and obesity was defined as BMI ≥ 30 kg/m 2 . 28 Place of upbringing included six categories of urbanization. Smoking behavior included current, former, and never smoker. More details about the questionnaire and definitions are available in Appendices, Questionnaire.
All citizens in Denmark are assigned a unique 10-digit personal identification number (civil registration number), which allows accurate linkage between individuals and register data. Data on age, sex, and current municipality residence were retrieved from Danish registers. Current municipality residence was divided into small (less than 100,000 citizens) and large density populated municipality (at least 100,000 citizens).
Current age was divided into 10-year age strata or two age groups (younger than 45 and 45 years or older). This upper age limit is normally set to exclude participants with chronic obstructive pulmonary disease, which is uncommon before the age of 45. 1,2,29 if the concentration of sIgE was ≥0.35 kU/L. 30 EDTA gel separated blood samples were centrifuged and stored at −20°C prior to testing. The blood-sampling date corresponded to the date of the questionnaire response.

| Statistical analysis
In the following, we define the prevalence of AR, AC, and asthma as the percentage who are currently or have a history of suffering from these conditions based on self-reported data.
Groups were compared by Mann-Whitney U test or Kruskal-Wallis test for non-normally distributed data, and chi-squared test for categorical data whenever all estimated counts were >5, and a 10 5 -fold Monte Carlo simulated estimate for Fisher's exact test whenever some estimated counts were ≤5. Predictors of risk were analyzed by multivariable logistic regression analysis (if two outcomes) and presented as odds ratios (ORs) with 95% CIs, or multinomial logistic regression analysis (if more than two outcomes) and presented as relative risk ratios (RRRs, the ratio between two risk estimates) with 95% CIs.
The association between time of birth (10-year birth cohorts) and age at the time of first symptom of asthma, AR, and AC in distinct single diseases and multimorbidity as well as in overlapping symptom groups were presented as cumulative incidence proportion and also analyzed by Poisson regression analysis. Participants entered the study at the date of birth and were right censored on the date of inclusion in the DBDS if they remained asymptomatic. Intervalcensored outcomes were constructed for participants experiencing symptoms of asthma, AR, and AC before inclusion. For multimorbidity groups, age at the time of first symptom of any of the current multimorbidities was used. In addition, the effect of the number of blood donations (1-10, 11-20, >20) on the association between birth year and age at the time of first symptom of asthma, AR, and AC were explored as a measure of survival bias. Results were presented as hazard ratios (HRs) with 95% CIs.
We tested for interactions in the Poisson and logistic regression models. If sex modified the effects of the exposures on the outcome, analyses were stratified by sex; otherwise, overall pooled estimates of the effects adjusted by sex were calculated.
The association between different Phadiatop cutoff values and self-reported ARC was analyzed by receiver operating characteristic (ROC) curve.  Oral and written informed consent was obtained from all participants.

| RESULTS
For demographics of the study population stratified by sex, see Table 1. Differences between males and females were present for allergic predisposition, current age, BMI, smoking behavior, place of upbringing, and the distribution between allergy and asthma phenotype groups.

| Prevalence of AR, AC, asthma, and inhalant allergen sensitization
The overall prevalence of self-reported AR, AC, and asthma among 52,976 blood donors was 19%, 15%, and 9%, respectively. The prevalence of inhalant allergen sensitization among 25,257 blood donors was 30%. Most cases of asthma (95%) were diagnosed by a doctor, 3.5% had current asthma, and 25% of the allergic participants reported asthma.
The cumulative incidence proportions of single diseases and multimorbidity were highest in the youngest birth cohorts, see Figure 1. We were not able to illustrate the cumulative incidence proportions for "asthma and AC" due to the low number of participants in this group. The cumulative incidence proportions of symptoms of AR, AC, and asthma regardless of multimorbidity were also highest in the youngest birth cohorts, se Figure A2. Stratifying by the total number of blood donations had no major influence on the effects of the birth year (Table S1), and the total number of blood donations was included as a confounder further on, see Tables 2 and A1. The youngest birth cohorts had the highest risk of experiencing "only asthma," "asthma and ARC," and "ARC," see Table 2. No associations between birth year and debut of symptoms were seen for those with either AR or AC. The youngest birth cohorts had also the highest risk of experiencing symptoms of asthma, AR, and AC regardless of multimorbidity and they experienced asthma symptoms at a significantly earlier age (p = 2.48 � 10 −26 ), see Figures A2 and S1.
Changing the cut-off value of a positive Phadiatop result 31 did not improve the overall performance of the test, see ROC curve analysis in Figure A3.

| Risk factors for AR, AC, and asthma
Differences between phenotype groups were present for inhalant allergen sensitization, allergic predisposition, current age, sex, smoking, and geographical residence, see Tables A2 and A3.
The multinomial logistic regression analysis confirmed that the inhalant allergen sensitized participants had a higher risk of having asthma, AR, AC and ARC than non-sensitized participants as compared to the controls, while adjusting for parental allergy, current municipality residence, obesity, smoking behavior, number of blood donations, and age (RRR range: 1.8-36.2 in females and 2.9-82.2 in males), see Table 3. Parental allergy also increased the risk of all phenotypes (RRR range: 1.7-3.8). Age younger than 45 increased the risk of non-allergic asthma, and obesity showed a trend of increased risk of asthma. Conversely, current smoking had a protective effect on ARC in male participants, see Tables 3, S2a and S2b.

| Associations between geographical upbringing and AR, AC, asthma, and allergen sensitization
Participants who grew up on farms with livestock had the lowest proportions of AR, AC, and inhalant allergen sensitization compared with the other urbanization categories, see Table A3 and A4 for characteristics of the study population stratified by phenotype group and inhalant allergen sensitization, respectively. Participants who grew up in inner cities had a higher proportion of parental allergy compared with those from farms with livestock (27% vs. 16%), and most of the participants from inner cities were residents in a large density populated municipality.
The multinomial logistic regression analysis among participants younger than 45 years showed that the protective effect of living on a farm with livestock during childhood was only present for acquiring ARC (with or without asthma) compared with almost all other urbanization categories, see Figure 2 and on that, ARC and inhalant allergen sensitization were further analyzed regardless of asthma and without age stratification. Farm childhood only had a protective effect on ARC in inhalant allergen sensitized participants, see Figure 3 and Table S4.

| Risk factors for inhalant allergen sensitization
The seroprevalence (inhalant allergen sensitization) was positively associated with male sex, allergic predisposition, age younger than 45 years, and living in a large density populated municipality, see

| Seasonal allergy and asthma
Most allergic participants had intermittent symptoms, and the months during which they usually experienced allergic symptoms were from April to August (corresponding to the hay fever season) with the highest proportion in June (grass pollen season in Denmark), see Figure A4. The proportions of participants with symptoms were generally higher in sensitized compared with non-sensitized participants. Persistent AR and AC symptoms were more frequent in participants with either AR or AC compared to those with ARC, as well as in sensitized compared with non-sensitized. Symptoms of both allergic and non-allergic asthma peaked during the winter; allergic asthma also peaked in the summer.

| Missing data
The number of unanswered questions in the questionnaire was approximately 4%. F I G U R E 1 Cumulative incidence proportions in single diseases and multimorbidity. Cumulative incidence proportions (and corresponding 95% confidence intervals) of self-reported asthma, allergic rhinitis (AR), and allergic conjunctivitis (AC) by age in 10-year birth cohorts in the overall study population, irrespective of sex. Stratified in distinct single diseases and multimorbidity groups. In multimorbidity, groups age refers to the age of first symptom irrespective of what disease. ARC, both AR and AC T A B L E 2 Association between birth cohorts and the age at the time of first symptom of asthma, AR or AC in single diseases and multimorbidity   Similar to other studies, our results showed increased risk of experiencing asthma, AR, or AC symptoms, in the youngest compared with the oldest birth cohort for participants with ARC and allergic asthma. However, contrary to other studies, this was also seen for asthma only. 36 In this study, we assessed both temporal and agerelated changes. Changes in symptoms and diseases may be related to changes in life style, 37 microbial exposure in the environment, 38 and changes in healthcare patterns, for example, changes in perception of illness and diagnostic practices over time.
A diverse and mature microbiota lowers the prevalence of atopy. [39][40][41][42] Hence, the urbanization categories were used as proxy measures for microbial load and diversity early in life. 1 Consistent with previous studies, growing up on farms (an environment rich in microbial load) was associated with lower risk of allergic sensitization, AR, and allergic asthma in adulthood. 2,10 Asthma, AR, and AC often coexist in the same individuals (multimorbidity). When stratifying our study population into two allergic groups, we found that farm childhood only had a protective effect on acquiring ARC (allergic multimorbidity) with and without asthma and not on acquiring either AR or AC. Combining self-reported data with data on inhalant allergen sensitization showed that the effect was only present in sensitized participants with ARC. We speculate that the protective effect of farm childhood on allergic disease in adulthood is explained by the protective effect on allergic sensitization which strongly correlates with ARC.
We further speculate that the parental allergy factor may also impact the results. A Swedish study showed that men with allergy more often chose to move from farms compared to those without allergy, resulting in a healthy worker effect with less atopy and less severe symptoms in the farming population. 43 In our study population, parental allergy was less frequent in participants from farms compared to participants from inner city and most of the participants from inner cities were currently residents in a large municipality (with higher exposure to air pollution). Nevertheless, adjusting for parental allergy and current municipality residence only slightly altered the results. Use of self-reported exposures and outcomes, in particular dating back to early life, may induce recall bias. However, we expect that participants were able to remember place of upbringing with a reasonable accuracy.
An association between obesity and asthma, especially, in females may exist. Though, the causality between obesity and asthma is still uncertain; a suggested explanation is due to changes in systemic metabolism. 15,44,45 The overall seroprevalence of inhalant allergen sensitization was 30% which is consistent with other studies. 9,10 As expected, inhalant allergen sensitization was positively associated with male sex, allergic predisposition, and younger age, 11,46,47 while current and former smoking had a protective effect. 48 The overall sensitivity of the Phadiatop analysis was 91% among participants with both ARC and asthma. This is in accordance with other studies showing that polysensitization is higher in individuals with allergic multimorbidity than in those with single diseases. [49][50][51] The Phadiatop analysis includes allergens from the nine most common respiratory sensitizers, but the participants could be sensitized to other allergens, have local AR, 52 or they could represent a non-allergic phenotype.
We found a strong association between allergic multimorbidity and asthma. It is, however, a limitation of our study that severe asthmatic individuals are not included. Allergic multimorbidity is tightly associated with severe asthma. 53-56 The absence of participants with severe asthma in our study population could cause underestimation of the prevalence of asthma with allergic multimorbidity (AR and AC) and explain why only 47% of our asthmatic participants have concomitant AR, while this proportion is 64.5% in another study using the ECRHS cohort. 57 The strength of our study is a large, overall healthy, and wellcharacterized study population. We included a detailed standardized questionnaire and an objective marker of allergic sensitization. The DBDS relies on the existing infrastructure of continuous collection of blood samples in the Danish blood centers and on the local staff at the blood centers, who collect questionnaire data during blood donation.
The participation rate was high, 24 and the number of unanswered questions in the questionnaire is low (approximately 3%). Prior to study inclusion, the questions were unknown to the blood donors. All participants in our study were unremunerated voluntary blood donors and had no incentive to non-disclosure of diseases. Although the crosssectional design hinders causal inference, it allows exploration of associations to generate hypotheses for further research.

| CONCLUSION
Birth year, childhood residence, inhalant allergen sensitization, and allergic predisposition had an effect on the risk of asthma, AR, and AC.
We found that individuals with self-reported ARC represent a primarily sIgE phenotype while those with either AR or AC represent more diverse phenotypes, and we suggest that the two phenotypes are analyzed separately in epidemiological studies based on only selfreported allergy.
The DBDS blood donor population proves well-suited for further studies of gene-environment interactions and biomarkers related to airborne allergy and asthma. In addition an affirmative answer to: "In which months of the year do this nose problem occur?" Months with nasal symptoms "How old were you when you first had hay fever or nasal allergy?" First nasal symptoms "When were the last time you had nasal symptoms?" Last nasal symptoms "Have you ever had a problem with sneezing, or a runny or a blocked nose when you did not have a cold or the flu?" Nasal symptoms "Do you have any eye allergies including hay fever?" Allergic conjunctivitis (AC) In addition an affirmative answer to: "In which months of the year do this eye problem occur?" Months with eye symptoms "How old were you when you first had hay fever or eye allergy?" First eye symptoms "When were the last time you had eye symptoms?" Last eye symptoms "Have you ever had a problem with itchy or watery eyes?" Eye symptoms "Do you have any nasal allergies including hay fever?" AND "Do you have any eye allergies including hay fever?" Allergic rhino-conjunctivitis (ARC) "Do you have or have you ever had asthma?" Asthma In addition an affirmative answer to: "Have you ever had asthma diagnosed by a doctor?" Doctor-diagnosed asthma Q u e s t i o n n a i r e (Continued)

Questions Definition
"Which months of the year do you usually have trouble with your breathing?" Months with asthma symptoms "How old were you when you had your first attack of asthma?" First asthma symptom At least one of the following:"Are you currently taking any medicine (including inhalers, aerosols or tablets) for asthma?""Have you been woken by an attack of coughing at any time in the last 12 months?""Have you been woken by an attack of shortness of breath at any time in the last 12 months?""When were the last time you had an attack of asthma?" -With the answer "Within the last 12 month".  symptom age 0-5 years were references. There were no interactions and overall pooled estimates of the effect were calculated. Significant differences after Bonferroni correction are shown in bold. Participants can appear in more than one symptom group.

of 23 -
T A B L E A2 Association between allergic rhino-conjunctivitis, asthma, and inhalant allergen sensitization (N = 25,257) Notes: Results are presented as percentages with corresponding 95% confidence intervals (CI) and medians with interquartile ranges (IQR). Groups were compared by Kruskall-Wallis test for non-normally distributed data, and chi-squared test for categorical data whenever all estimated counts were >5, and a 10 5 -fold Monte Carlo simulated estimate for Fisher's exact test whenever estimated counts were ≤ 5. P-values were Bonferroni corrected, *p < 0.05; **p < 0.001; NS: non-significant.

Self-reported diseases and symptoms Sex
Abbreviations: ARC, allergic rhinitis (AR) and allergic conjunctivitis (AC); F, female; M, male; non-sensitized, allergen-specific immunoglobulin E (Phadiatop) <0.35 kU/L; sensitized, allergen-specific immunoglobulin E (Phadiatop) ≥ 0.35 kU/L. a Compared with the total number of participants in the column. b Compared with the total number of participants in the row.

F I G U R E A 2
Cumulative incidence proportions in each of the birth cohorts. Cumulative incidence proportions (and corresponding 95% confidence intervals) of self-reported asthma, allergic rhinitis (AR), and allergic conjunctivitis (AC) by age in ten-year birth cohorts in the overall study population, irrespective of sex. Participants can appear in more than one symptom group MIKKELSEN ET AL.
-21 of 23 F I G U R E A 3 ROC curve analysis. Association between different Phadiatop cutoff values and self-reported allergic rhinitis and allergic conjunctivitis (ARC) analyzed by receiver operating characteristic (ROC) curve with reported area under the curve (AUC)

F I G U R E A 4
Months with symptoms of allergy and/or asthma. Nasal-, eye-or asthma symptoms in participants with allergic rhinitis (AR), allergic conjunctivitis (AC), and asthma (n = 6278) stratified by inhalant allergen sensitization. Results are presented as proportions of the allergic and/or asthmatic participants. The figure shows both sexes in one. Participants could give more than one answer and appear in more than one symptom group