Sociodemographic factors linked to food allergy diagnosis among high-risk children with atopic dermatitis

Abstract

Background:

Atopic dermatitis (AD) is a known risk factor for the development of food allergy (FA). Prior work has suggested disparities in diagnosis/management of FA in urban populations.

Objective:

To determine whether socioeconomic conditions, as measured by the area deprivation index and insurance status, or racial/ethnic self-identity was associated with risk of FA diagnosis (DFA), evaluation by an allergist, or objective FA testing among high-risk children with AD.

Methods:

This is a retrospective cohort study of pediatric patients with physician-diagnosed AD who had received primary care at a single urban academic tertiary care center between 2009 and 2022. Statistical analysis in SPSS (IBM Corp. Released 2017. IBM SPSS Statistics for Windows, Version 25.0, Armonk, NY) used χ², analysis of variance, and logistic regression.

Results:

In a total of 3365 pediatric subjects, 41.3% identified as non-Hispanic Black, 33.9% Hispanic, 6.9% Asian, and 14.9% non-Hispanic White. Hispanic children with AD and DFA were significantly less likely to be evaluated by an allergist than White or Asian children (65.9% vs 82.8% and 80.3%, P = .001 and P = .02). Non-Hispanic Black children with AD and DFA were more likely to have no objective FA testing than White children (20.9% vs 12.1%, P = .04). The White and Asian children were more likely to undergo the thorough combination of both blood and skin testing for DFA than Black or Hispanic children (15.5% and 22.4% vs 7.1% and 7.9%, respectively, P = .007, P = .00005, P = .03, P = .0008).

Conclusion:

Labeling at-risk young children with FA without thorough objective testing can affect their nutrition and quality of life. Barriers to equitable evaluation of DFA should be further investigated and addressed.

Introduction

Food allergy (FA) is a common condition affecting a significant proportion of the US pediatric population with a marked burden on children’s lives and well-being, especially among children of historically underrepresented populations and those affected by lower socioeconomic conditions.^1,2 One study revealed that Black and Hispanic children were approximately twice as likely as White children to be found in the emergency department for food-induced reactions (34%−39% vs 18%) and to have higher rates of food-induced anaphylaxis (33%−35% vs 16%).³ FA is strongly linked to another atopic condition, atopic dermatitis (AD), which also disproportionately affects Black and Hispanic children.^4,5 It is noteworthy that AD is a major risk factor for FA and often precedes its development in childhood.⁶ One study revealed that infants with AD were 6 times more likely to have egg allergy and 11 times more likely to have peanut allergy at 12 months of age than infants without AD.⁷ A proposed mechanism for this link is increased passage of food antigens through inflamed skin due to barrier dysfunction of eczematous skin, subsequently leading to sensitization and development of intolerance against food proteins.⁸

However, why some children with AD will ultimately be diagnosed with FA and some will not is not fully clear. As mentioned previously, both FA and AD affect Black and Hispanic children disproportionately. Furthermore, studies have revealed that many of these related disparities are due to sociodemographic disadvantages that affect children of historically underrepresented racial/ethnic groups.^3,9–11 This brings up the question of whether there are true risks within certain racial, ethnic, or socioeconomic groups of developing a FA or whether these populations tend to be burdened with underdiagnosis or overdiagnosis of FA. Overdiagnosis of FA may ultimately be harmful to these groups, as it has been found through studies that early consistent exposure to common food allergens reduces the incidence of developing FA.¹²

Our current study aimed to investigate whether race, ethnicity, and socioeconomic conditions of children with AD affect the presence of physician-diagnosed FA within the study population. We also aimed to discover whether physician diagnoses of FA are supported with objective data across all patient groups, specifically with FA testing and referrals to an allergy specialist. Our study uniquely used a validated proxy called the area deprivation index (ADI), which incorporates various factors within the domains of income, education, employment, and housing quality to measure and rank the socioeconomic conditions of the study subjects’ home neighborhoods.¹³

Methods

Study Population

Our study included pediatric patients aged 0 to 18 years old with a physician diagnosis of AD, with the presence of an International Classification of Diseases, 10th Revision (ICD-10), diagnosis code and confirmed through chart review of the electronic medical record (EMR) documentation. All patients had been seen at our institution for at least 1 primary care visit between 2009 (when the EMR was implemented) and February 8, 2022, when initial data extraction occurred. Subjects seen exclusively for emergency department visits, inpatient admissions, or specialist appointments were excluded from data analysis.

Data Collection

Using the above-mentioned criteria, we developed a retrospective cohort of subjects and performed detailed chart review for the duration of all visible records in the EMR. RedCap was used for secured data collection and organization (REDCap Version 13.6.1. Released 2024, Vanderbilt University.). The study was approved by the institutional review board.

On initial data extraction, we collected birth date, sex (male or female as documented in the medical record), self-identified race and ethnicity as documented in the EMR, body mass index (BMI), current patient address listed in EMR, and insurance provider/plan name. Race and ethnicity were combined into 1 reported classification: non-Hispanic White, non-Hispanic Black, Hispanic/Latinx, Asian, or Other/Unknown. We also extracted each subject’s allergy list and problem list to preliminarily determine whether they carry a FA diagnosis (DFA).

Our team of 7 researchers then completed independent retrospective chart reviews for each subject to gather and confirm specified quantitative and qualitative data. We were able to view all electronic records from our institution and those from most regional outside institutions using the CareEverywhere platform within Epic (Epic Systems, Verona, WI. Customized for Rush University Medical Center in Chicago, IL, Last updated 2024).

If the patient had a documented DFA, further chart review was pursued for detailed diagnostic factors. DFA was defined as an ICD-10 code, allergy list entry, and/or problem list entry written by a physician in the patient’s chart. The focus was on the presence of specific DFA to egg, milk, wheat, soy, peanut, tree nut, fish, shellfish, and sesame. If an allergy to another food was present in a patient’s chart, it was collected as other. Inherent in this method, we included all diagnoses of “food allergy” documented by physicians for these children, but it is possible that this could inaccurately include lactose intolerance, celiac disease, or other non−IgE-mediated food reactions. Thus, for each specific DFA history, it was noted whether there were objective FA test results confirming the diagnosis, and if so, which type of test was completed: percutaneous, blood, oral challenge, and/or unknown (test done in past/outside facility). The largest documented wheal size (in millimeter) from skin prick testing and/or highest specific IgE value from serum testing were recorded for each food. This was a cross-sectional study, in that we were noting which patients carry current FA diagnoses at the time of data collection, rather than analyzing longitudinally for the development of new FA diagnoses or development of tolerance to foods which previously caused allergic reactions.

We also determined whether each subject had seen an allergist/immunologist at our institution or an outside institution. We recorded the first and last dates of allergist/immunologist clinic visits and the number of visits during that period. In our electronic charts, we can track referral orders placed to allergists within our institution, or to external providers, and we recorded these data.

The principal reviewer (primary author) personally instructed all the other data collectors in the chart review process. This author personally reviewed 65% of the charts. The charts reviewed by other authors were double checked for accuracy by the primary author.

Determining Socioeconomic Status and Health Disadvantage

For our study, the subject’s addresses were matched to their census block group, which is more precise than zip code and is used to define “neighborhood” for the purposes of ADI tabulation. ADI determining factors included theoretical domains of income, education, employment, crime, and housing quality. We chose to report the ADI for each patient as the national percentile ranking from 1 to 100 provided in the most recently available 2019 database. A higher ADI value indicates more socioeconomic disadvantage in the neighborhood, the highest being 100, and a lower ADI value indicates less socioeconomic disadvantage, with lowest being 0.

Insurance status was also collected, specifically whether insurance was private or public (Medicaid). Medicaid has been found to be a reliable proxy for family’s economic status.¹⁴ These data points were used as an additional tool to interpret socioeconomic conditions.

Data Analysis

The data set was thoroughly checked and adjusted for missing data and outliers. The data were analyzed using the statistical software IBM-SPSS 25.0 (IBM Corp. Released 2017. IBM SPSS Statistics for Windows, Version 25.0. Armonk, NY: IBM Corp.) Furthermore, χ² and analysis of variance were initially used to test significance of observed differences between the groups. Logistic regression analysis was then used to evaluate the association of various presumed risk factors (age, sex, BMI, race/ethnicity, insurance status, ADI, and whether patient had seen an allergist) with the diagnosis of top 9 FA specifically, along with other secondary analyses.

Results

A total of 3365 pediatric patients who had a physician diagnosis of AD and received primary care at our large academic tertiary care center within the city of Chicago were identified for inclusion in this study. In this population, 41.3% identified as non-Hispanic Black, 33.9% Hispanic, 6.9% Asian, and 14.9% non-Hispanic White per the EMR. There are 99 subjects classified as “Other” (3%). This includes those with Other or Refusal listed as race/ethnicity in the EMR, including 4 patients who identified as American Indian or Native American, a population too small for statistical analysis. Most of the study subjects had Medicaid insurance; 61.4% were covered by Medicaid vs 38.5% covered by private insurance. Among the study population, 756 patients (22.5%) have been diagnosed with 1 or more top 9 FAs (egg, milk, wheat, soy, peanut, tree nuts, fish, shellfish, or sesame), which will henceforth be referred to as DFA.

Non-Hispanic Black and Hispanic patients were significantly more frequently covered by Medicaid (76.3% and 62.8% respectively, as compared with 37.5% of Asian subjects and 29.1% of non-Hispanic White subjects; P < .001). As depicted in Figure 1, subjects of historically underrepresented race/ethnicities (non-Hispanic Black and Hispanic) tended to live in neighborhoods with higher ADI, which is indicative of greater socioeconomic disadvantage.

Figure 1.

Non-Hispanic Black and Hispanic children in this study population (large, urban, high representation of traditionally underrepresented racial/ethnic groups in medical research) tend to live in neighborhoods with more socioeconomic disadvantage than non-Hispanic White and Asian children in the population.

Diagnosis of Food Allergy

As displayed in Table 1, there were several significant differences between the group with DFA and the group with no DFA. A significantly higher percentage of male subjects had DFA than female subjects (24.2% vs 20.6%, P = .01). There were differences in the presence of DFA by race/ethnicity: 32.8% of Asian subjects, 24.4% of non-Hispanic Black subjects, 23.1% of non-Hispanic White subjects, and 17.1% of Hispanic subjects affected. Asian subjects were statistically significantly more likely to have a DFA than the other groups, but other observed differences were not statistically significant.

Table 1.

Diagnosis of Food Allergy to 1 or More of the Top 9 Food Allergens Among 3365 Children With Atopic Dermatitis Diagnosed at a Single Tertiary Academic Center

Sample characteristic Total population = 3365	Diagnosis of top 9 food allergy, n (%) (n = 756)	No DFA, n (%) (n = 2609)	Comparison ofDFA and no DFA groups by sample characteristic (unadjusted P value)	Adjusted oddsof top 9 DFA (coefficient with 95% CI)
Sex			.012
Female	334 (44.2)	1290 (49.4)		0.80 (0.68–0.94)
Male	422 (55.8)	1319 (50.6)		Ref.
Age			.078	1.03 (1.01–1.05)
<2 y	78 (10.3)	299 (11.4)		(Age as a continuous variable for this logistic regression)
2–8 y	235 (31.2)	900 (34.5)
8–14 y	321 (43.3)	1003 (38.4)
>14 y	122 (16.1)	407 (4.7)
BMI (mean ± SD)	19.80 ± 5.86	19.89 ± 6.21	.72	0.99 (0.98–1.01)
Race/ethnicity			<.001 a
Non-Hispanic White	116 (23.1)	386 (76.9)		Ref.
Non-Hispanic Black	340 (24.4)	1051 (75.6)		1.42 (1.09–1.87)
Hispanic	205 (17.8)	936 (82.2)		0.89 (0.68–1.12)
Asian	76 (32.8)	156 (67.2)		1.68 (1.18–2.38)
Other	19 (19.2)	80 (80.8)		0.91 (0.53–1.57)
Insurance type			.001
Medicaid (public)	426 (20.6)	1641 (79.4)		Ref.
Private	328 (25.3)	966 (74.7)		1.27 (1.06–1.53)
ADI (mean ± SD)	46.1 ± 21.82	49.2 ± 20.92	<.001	0.994 (0.990–0.998)

Abbreviations: ADI, area deprivation index; BMI, body mass index; DFA, food allergy diagnosis; Ref., reference.

NOTE. This logistic regression was adjusted for sex, age, BMI, race/ethnicity, insurance type, and ADI. Top 9 food allergy includes egg, milk, wheat, soy, peanut, tree nuts, fish, shellfish, and sesame. Age is defined as the age of subject at time of data extraction. Owing to some missing data, the numbers in all rows might not add up to the total.

^aHispanic children had lower rate of food allergy compared with all other groups. Asian children had higher rate of food allergy diagnosis compared with all other groups.

We performed logistic regression analyses incorporating multiple factors that may be associated with DFA: age, sex, BMI, racial/ethnic identity, ADI, and insurance status (see adjusted odds column in Table 1). When controlling for these factors, non-Hispanic Black and Asian children were more likely than non-Hispanic White children to have DFA (Black: odds ratio [OR] 1.42, 95% CI: 1.09–1.87; Asian: OR 1.68, 95% CI: 1.18–2.38). Children with private insurance were significantly more likely to have DFA than patients with public/Medicaid insurance (OR 1.27, 95% CI: 1.06–1.53, P < .01). The mean ADI was significantly lower (indicative of less socioeconomic disadvantage) in the group with DFA as compared with those without DFA (OR 0.994, 95% CI: 0.990–0.998, P < .001). Older patients were also more likely to have DFA (coefficient 1.04, P < .01).

Evaluation by Allergy/Immunology Specialist

Among the study population of children with AD, 1001 were evaluated by an allergist/immunologist. Table 2 compares children who were seen by an allergist specialist vs those who were not; logistic regression was adjusted for race/ethnicity, sex, ADI, insurance type, and presence of atopic diagnoses (FA, asthma, and allergic rhinitis), which are factors that may affect likelihood of being seen by an allergist. Asian children were more likely to have seen an allergist compared with all other race/ethnicity groups. Sex was not associated with allergist visits. Children with private insurance (non-Medicaid) were significantly more likely to have seen an allergist than those covered by Medicaid (34.0% vs 27.1%, unadjusted P < .001). Subjects who had been evaluated by an allergist were significantly more likely to have DFA or diagnosis of asthma or allergic rhinitis. ADI was not significantly associated with seeing an allergist in the adjusted model.

Table 2.

Differences Among Pediatric Patients With Atopic Dermatitis Seen by an Allergist vs Those Who Are Not

Sample characteristic Total population = 3365	Seen by an allergist, n(%)(n = 1001)	Never seen by an allergist, n(%)(n = 2364)	Adjusted odds ofbeing seen by an allergist (odds ratio is displayed with 95% CI)
Sex
Male	557 (32.0)	1184 (68.0)	Ref.
Female	444 (27.3)	1180 (72.7)	0.948 (0.788–1.139)
Race/ethnicity
Non-Hispanic White	168 (33.5)	334 (66.5)	Ref.
Non-Hispanic Black	425 (30.6)	966 (69.4)	0.930 (0.690–1.252)
Hispanic	274 (24.0)	867 (76.0)	0.766 (0.570–1.028)
Asian	101 (43.5)	131 (56.5)	1.707 (1.146–2.542)
Other	33 (33.3)	66 (66.7)	1.444 (0.834–2.502)
Insurance type
Medicaid (public)	560 (27.1)	1507 (72.9)	Ref.
Private	441 (34.0)	857 (66.0)	1.312 (1.069–1.610)
ADI (mean ± SD)	45.7 ± 21.9	49.7 ± 20.8	0.996 (0.991–1.000)
Top 9 food allergy diagnosis
Yes	562 (74.3)	194 (25.7)	Ref.
No	439 (16.8)	2170 (83.2)	0.091 (0.074–0.112)
Asthma diagnosis
Yes (% of cases in each the column that were diagnosed with asthma)	433 (46.0)	509 (54.0)	Ref.
No	567 (23.4)	1854 (76.6)	0.668 (0.542–0.823)
Allergic rhinitis diagnosis
Yes	673 (50.7)	655 (49.3)	Ref.
No	327 (16.1)	1709 (83.9)	0.264 (0.218–0.321)

Abbreviations: ADI, area deprivation index; Ref., reference.

NOTE. Comparisons were performed by binary logistic regression comparing children seen by an allergist with those who have not seen an allergist adjusted for sex, race/ethnicity, ADI, insurance type, and presence of food allergy/asthma/allergic rhinitis diagnoses.

Among the subgroup of children with DFA, those of non-Hispanic Black and Hispanic racial/ethnic identities were less likely to have been evaluated by an allergist, as found in Figure 2A. We observed that 82.8% of non-Hispanic White children and 80.3% of Asian children with AD and DFA had been evaluated by an allergist, as compared with only 74.7% of non-Hispanic Black children and 65.9% of Hispanic children. The lower likelihood of allergist evaluation compared with White children had a trend toward significance in Black children (P = .07) but was significant for Hispanic children in the adjusted model (P < .05). Children with DFA who were seen by an allergist lived in neighborhoods with a significantly lower ADI (45.1 vs 49.1, P = .02) and were significantly less likely to have Medicaid insurance coverage (53.5% vs 64.4%, P = .02).

Figure 2.

Lack of allergist specialist evaluation among children with AD and DFA. (A) Both non-Hispanic Black and Hispanic/Latino children with AD and DFA were less likely to be evaluated by an allergist than non-Hispanic White and Asian children. This difference was statistically significant for Hispanic children but not for Black children. (B) Many children with AD and DFA who were never seen by an allergist did have a referral in place to Allergy. The non-Hispanic Black children never seen by an allergist were significantly more likely to have an allergy referral in place than the non-Hispanic White children never seen by an allergist (P = .02). Other observed differences were not statistically significant. χ² analysis followed by custom tables for post hoc pairwise comparisons was used for demonstrated differences among all defined race/ethnicity groups. Furthermore, multiple logistic regression analyses were conducted, which revealed the reported differences remained significant after adjusting for sex, ADI, and insurance type. AD, atopic dermatitis; ADI, area deprivation index; DFA, food allergy diagnosis.

Referral for Specialist

We then evaluated if the children who had not been seen by an allergist were ever given a referral to an allergist specialist. First, we noted that significantly more Black and Hispanic patients (83.5% and 73%, respectively) in the entire study population had a referral in place before seeing an allergist, as compared with White and Asian children (56.0% and 64.4%, respectively). It is known that Medicaid-insured patients typically require a referral before seeing a specialist. As found in Figure 2B, we then analyzed the children with AD and DFA who had never been seen by an allergist. Many of these children did have an allergist referral in place but had never come to an allergy clinic: 53.4% of Black children, 38.6% of Hispanic children, 25% of White children, and 46.7% of Asian children.

For the subgroup of children with DFA, we performed logistic regression with a model incorporating age, sex, BMI, race/ethnicity, ADI, and insurance status to understand potential factors underlying the likelihood of referral to an allergist specialist. Sex, ADI, and BMI were not significantly associated with having a referral placed to Allergy. Among race/ethnicity groups, Black children had a significantly higher likelihood of having a referral placed to Allergy (OR 2.996, 95% CI: 1.792–5.007). Private insurance status was associated with a significantly lower likelihood of having a referral placed to Allergy (OR 0.502, 95% CI: 0.350–0.719). Higher age was also associated with a significantly lower likelihood of having a referral placed to Allergy (OR 0.938, 95% CI: 0.902–0.976).

Confirming Food Allergy With Testing

We attempted to investigate whether subjects with DFA had undergone documented objective FA testing. As displayed in Figure 3, Black children with AD and DFA were significantly more likely than White children with AD and DFA to have no objective testing to support the DFA (20.9% vs 12.1%, P = .04). In this study population, all children with AD and DFA who underwent objective testing of FA lived in neighborhoods with lower mean ADI (higher socioeconomic conditions) than those children with no objective testing (45.1 vs 49.7, P = .02). Children who did not undergo any objective testing were more likely to have Medicaid insurance coverage than children who underwent objective testing to evaluate their FA diagnoses (54.5% vs 65.0%, P = .02).

Figure 3.

Non-Hispanic Black children with atopic dermatitis and a food allergy diagnosis were significantly more likely than non-Hispanic White children to have no objective testing to support that food allergy diagnosis and less likely than both non-Hispanic and Asian children to have undergone both skin and blood testing. Hispanic/Latino children with atopic dermatitis and a food allergy diagnosis were significantly less likely than non-Hispanic White and Asian children to have undergone both skin and blood testing. χ² analysis followed by Custom Tables for post hoc pairwise comparisons was used for demonstrated differences among all defined race/ethnicity groups. Furthermore, multiple logistic regression analyses were conducted, which revealed the reported differences remained significant after adjusting for sex, ADI, and insurance type. ADI, area deprivation index.

Black children were also significantly less likely than White or Asian children with AD and DFA to have undergone both skin and blood testing to evaluate for DFA (7.1% vs 15.5% and 22.4% respectively, P < .01 and P < .001, respectively). Hispanic children were also significantly less likely than White or Asian children to have undergone both skin and blood testing (7.9% vs 15.5% and 22.4% respectively, P < .05 and P < .001, respectively). Of note, subjects with “negative testing” had test results without any elevated specific IgE or a skin test reaction to the food but still carried a DFA in the medical record.

Discussion

Our study offers the unique perspective of characterizing the association between AD and DFA in the real-world clinical setting among a large population of pediatric patients already diagnosed with AD rather than the general population. Compared with prior studies, this work offers a different perspective with its analysis of physician-documented atopic disease diagnoses in the medical record (rather than parent-reported survey analyses) and a high representation of historically underrepresented racial/ethnic identities. By including only children receiving primary care within the academic institution, we were able to provide an analysis of allergist specialist referral, evaluation, and objective testing patterns for those with DFA.

We first revealed that the non-Hispanic Black and Hispanic children in this study were significantly more likely to be covered by Medicaid/public insurance and tended to live in areas with higher socioeconomic disadvantage (higher ADI) than their White or Asian peers. This matches the known disparities affecting historically underrepresented populations in the United States, especially in inner-city locations like where this study was conducted.¹⁴ Non-Hispanic Black and Hispanic children with AD and DFA were found to be less likely to see an allergist than non-Hispanic White and Asian children. A prior study by Taylor-Black and Wang¹⁵ found no significant difference in likelihood of allergist referral by race among a population of general pediatric patients with documented FA (67% were referred to an allergist, and 45% were actually evaluated by an allergist). In our study population, we looked specifically at the children never evaluated by an allergist and found higher referral rates for non-Hispanic Black and Hispanic children, which is most likely due to the requirements of the state Medicaid insurance plans these populations more frequently were covered by. Although many privately insured patients do not need a primary care referral to see a specialist, Medicaid does require such referral.

However, irrespective of being referred to an allergist, in our series being evaluated by an allergist was associated with lower ADI and having private insurance. Similar findings have been reported in children with asthma. Children with asthma insured by Medicaid are less likely to receive care from subspecialists for asthma.¹⁶ This is important especially because DFA was higher among AD children who were evaluated by an allergist. Furthermore, an allergist evaluation was associated with higher rate of asthma and AR diagnoses in the adjusted models. This finding raised the question whether the link between allergist evaluation and allergic diagnoses (DFA, asthma, and allergic rhinitis) is due to the allergic conditions that prompted the evaluation, or it is vice versa that thorough evaluation by allergists made these diagnoses. It could also be the combination of these factors that results in the observed association. To answer this question, we analyzed the subgroup with an allergist referral in the system. The association between allergist evaluation and allergic diagnoses stayed statistically significant in the subgroup of patients who were referred to an allergist. Furthermore, the association of private insurance and lower ADI (more affluent) with higher rate of DFA is in line with these findings, indicating access to care has resulted in allergic conditions being diagnosed. These all indicate that there are mostly likely several factors contributing to these associations, and it highlights the need and importance of specialist evaluation for children with AD regardless of having current symptoms of other atopic conditions.

There are multiple barriers to care that could be considered to explain these findings. As explained previously, although Black and Hispanic children with AD and DFA are less likely to see an allergist in clinic, many of them did have a referral order placed by the primary care provider. Therefore, a referral in the system—although an important step—does not solve the issue of access. There are important logistic and financial barriers to presenting for an allergist evaluation. Time off work, childcare, and transportation are required to access the clinic.¹⁷ Lack of a personal car and poor reliability of public transportation have previously been reported as specific barriers for low-income urban children who have missed scheduled medical appointments.¹⁸ Families may also anticipate high costs of evaluation and be reluctant to attend the visit. The cost of performing skin testing, blood testing, or oral food challenges and the required clinic staffing may also affect whether physicians offer these tests and whether families are interested in pursuing these tests. It is also probable that there is a lack of understanding and knowledge about allergies, available allergy testing, and utility of such an evaluation.

In understanding objective testing of FA, we found that non-Hispanic Black patients were more likely to have DFA without any objective testing. Although this data point may provide potential reassuring evidence against underdiagnosis of FA, it indicates possible overdiagnosis. Skin testing and oral challenge can be very important to clarify potential FA identified on panel in vitro testing (which was the sole testing method for many children in our study). As depicted in Figure 3, we did find that non-Hispanic White and Asian children in our population were more likely to undergo multiple types of objective testing for their DFA (both skin and blood testing) when compared with Black or Hispanic children. Additional subgroup analysis of children who were evaluated by an allergist found no difference in the number of children who underwent both skin and blood testing across all racial/ethnic identity groups (data not revealed). This analysis points away from unconscious bias of allergists as a reason for the observed differences in Figure 3 and instead indicates that the differences found among the whole group are more likely due to lack of complete workup and access to allergist specialty care. Access to an allergy clinic will be essential for ensuring that children and families are not burdened with unnecessary dietary restrictions or mislabeled to several food allergies that could affect lifestyle, mental health, and nutritional status.¹⁹ There may also be significantly higher associated cost of obtaining allergen-free foods because of an inaccurately diagnosed FA. Another important factor to consider is that mislabeling a young at-risk child with FA, and the resultant avoidance of the food can increase the risk of true IgE-mediated allergy in these children.

More attention has been appropriately dedicated in recent years to acknowledging systemic racism in the United States and its far-reaching impacts on the health of many non-White Americans, particularly those with a non-Hispanic Black racial identity. Our adjusted data reveal that the rate of DFA is not different among racial/ethnic groups once adjusted by socioeconomic groups indicating there is no link between race as a social construct with DFA once adjusted accordingly. Nevertheless, DFA was linked to the type of insurance and ADI indicative of its link to barriers and access to care.

There are several key limitations of this study to note. Its retrospective nature means that analysis was limited to information directly accessible in the EMR. It is possible that we missed allergist evaluations, objective test results, and referral orders in other health care systems. With the use of the Epic software and its CareEverywhere feature, we were able to access outside records for many local Allergy/Immunology clinics, which strengthened this study. We used each subject’s single current documented home address to tabulate ADI, which could be misleading if patients were lost to follow-up from our system or had lived in multiple neighborhoods with different ADIs. The segregation and socioeconomic stratification of the city of Chicago might make this less likely in our particular urban environment. We also had 99 patients with no documented racial/ethnic identity and did not analyze this “Other/Unknown” group, including the small number of patients who identified as Native Americans. In addition, a thorough clinical history is essential to accurate diagnosis of FA, but we were not able to measure the quality of the FA history as part of this particular study.

Despite its limitations, this study adds a novel perspective to the current literature specific to children with AD, who already have 1 atopic disease and are at risk for FA. Furthermore, this study used ICD-10 diagnoses documented in the chart by physicians and not only patient-/parent-reported history. The study also provides further evidence of racial/ethnic and socioeconomic disparities in DFA. These data indicate a need for prospective studies and potential interventions, such as an educational initiative at the primary care clinic level, to encourage ultimate access to allergist evaluation. Given the recent emphasis on addressing racial inequity in our country and the recent observed increased prevalence in atopic diseases, now is an essential time to further characterize these disparities so that we can mitigate underlying barriers and aim for equitable care for all children with atopy.