Food Allergy Reaction Severity and Management in a Diverse Population

Abstract

Background:

Definitive treatment for food allergy reactions including anaphylaxis varies widely by reaction severity and socioeconomic status but little data exists to characterize the relationship between severity, management, and race and ethnicity.

Objective:

We aimed to analyze the differences in reaction severity, epinephrine use, and emergency room utilization by race and ethnicity in a large, diverse, food-allergic cohort.

Methods:

We analyzed intake data from participants in the FORWARD cohort on the history of food allergy reactions, severity of the reactions, and management associated with each reaction. We used descriptive statistics as well as mixed-effects logistic and poisson models to describe the differences in reaction severity, ER visits, and total lifetime epinephrine usage by race and ethnicity.

Results:

784 children were included in the analysis: 425 (54.2%) were non-Hispanic White (NHW), 282 (36.0%) were non-Hispanic Black (NHB), and 77 (9.8%) were Hispanic or Latino (H/L). NHB children had increased odds of more severe reactions (OR: 1.7, 95% CI: 1.2, 2.3) and higher odds of going to the ER (OR: 2.8, 95% CI: 1.4, 5.4)). Both NHB (IRR: 0.4, 95% CI: 0.3, 0.5) and H/L (IRR: 0.3, 95% CI: 0.2, 0.5) children had lower rates of total lifetime epinephrine use.

Conclusion:

There are significant disparities in the severity and treatment of food allergy reactions by race and ethnicity, resulting in increased ER utilization and decreased total lifetime epinephrine. Equipping parents with resources and tools on management of food allergy reactions may result in decreased disparity in access to definitive care.

Introduction

Food allergy affects a significant portion of the pediatric population. Within the United States, 8% of children have a parent-reported food allergy¹. Reactions following exposure to food allergens can vary greatly in severity and symptomatology, and those that progress to severe anaphylactic reactions often begin initially with mild symptoms². Anaphylaxis is a medical emergency and requires rapid intervention given the potential for life-threatening symptoms^3–4. The recommended first line treatment for anaphylaxis is epinephrine, and administration of an Epinephrine Auto-Injector (EAI) at the first sign of symptoms is the gold-standard treatment which has been demonstrated to prevent progression in reaction severity and decrease rates of anaphylaxis-associated hospitalizations^4–5. Furthermore, delays in epinephrine administration have been shown to increase the risk of progressing to severe reaction following exposure to food allergens⁴.

Despite the importance of prompt EAI administration for anaphylaxis, current rates of daily EAI carriage and prompt EAI usage in patients with food allergy and anaphylaxis have been reported to be inadequate.⁶ The explanation for inadequate EAI carriage and usage is multifactorial. Recent research has shown that socioeconomic factors have an impact on the quality of food allergy management, including education on reaction management and reliable access to EAI usage⁷. Patients of any age who have a history of filling their epinephrine prescriptions and have previously been seen by an allergist/immunologist are known to have lower risk for severe reactions⁷. These protective factors are correlated with socioeconomic status, as prior data has shown that children in U.S. counties with higher poverty rates have lower rates of filling prescriptions for EAI and lower rates of visiting an allergist/immunologist following initial food allergy diagnosis.⁸ Children in impoverished counties also visit the ER for food allergy at higher rates than those in more affluent areas, further suggesting food allergy management remains inadequate for those with fewer financial resources.⁸

In addition to these documented socioeconomic differences, racial disparities have independently been shown to affect food allergy management and outcomes.⁹ Analysis of data from the US National Mortality Database collected from 1999 to 2010 revealed that, when compared to non-Hispanic White populations, non-Hispanic Black populations are at greater risk for fatal food-induced anaphylactic reactions.¹⁰ In order to improve outcomes and minimize racial disparities in pediatric patients with food allergy, it is crucial to better understand the factors influencing perceived severity and treatment outcome disparities. With this study, we aim to investigate and compare characteristics of patient-reported food allergy reaction severity, EAI usage, and emergency room utilization amongst non-Hispanic white (NHW), non-Hispanic black (NHB), and Hispanic/Latino (H/L) children with known food allergy.

Methods

The Food Allergy Outcomes Related to White and African American Racial Differences (FORWARD) study is a prospective, multi-site longitudinal cohort study designed to characterize the natural history of pediatric FA among a diverse cohort of children. Beginning in 2017, we enrolled children ages 12 and under with a physician-diagnosed FA from four clinical sites: Anne & Robert H Lurie Children’s Hospital and Rush Children’s Hospital in Chicago, Cincinnati Children’s Hospital in Cincinnati, and Children’s National Hospital in Washington D.C. Further details regarding the design of the FORWARD study have been previously described.

The data analyzed from this study originate from the intake survey of the FORWARD study, administered at the time of enrollment. This instrument captured characteristics of the most severe food allergy reactions occurring by the time of enrollment. Respondents characterized symptoms associated with the most severe food allergy reaction for each current food allergy. The age at first reaction was computed as the age at which the participant first experienced a food-allergy reaction across all foods to which they were allergic. The severity of each food allergy reaction reported at intake was rated by the parents as either Mild, Moderate, Severe, and Life was in Danger. Parents were also surveyed on the total lifetime epinephrine use across all reactions for their child.

Summary and descriptive statistics were calculated using Pearson’s chi-squared tests for comparisons of categorical outcomes by race and ethnicity and ANOVAs were constructed to compare continuous outcomes. A mixed-effects ordinal logistic regression model was used to model the odds of a 1-level increase in reaction severity while holding constant the variables of race and ethnicity, age at first reaction, age at enrollment, sex, and enrollment site, with a random intercept and slope for each participant for Study ID to account for between-reaction, within-subject correlation. A similar mixed-effects logistic model was fit for the outcome of treating a food allergy reaction with visiting the ER additionally adjusted for reaction severity and receiving epinephrine for that reaction. Finally, a Poisson model was fitted to model the incidence ratio of total lifetime epinephrine usage while adjusting for the previously mentioned variables in addition to specifying age at enrollment as an offset, including having ever utilized the ER during a reaction and the maximum reaction severity experienced by the subject across all reactions reported by intake.

Results:

In total, 784 children were included in the analysis (Table 1). 425 (54.2%) children were NHW, 282 (36.0%) were NHB, and 77 (9.8%) were H/L. The majority of children were male (61.7%) and this ratio did not significantly differ between the three groups. On average, NHB children were older compared to NHW children (7.0 vs. 5.3, p < 0.001) and were also older, on average, at their first reaction (20.6 months vs. 13.3 months, p < 0.001). Across the cohort, the average number of distinct food allergies was 3.3 (SD: 1.8). The data suggested that NHB children had more reactions on average than NHW and H/L children (3.6 vs. 3.2 vs. 2.9, p = 0.009). Additionally, NHB and H/L children more frequently reported severe reactions compared to NHW children (NHB: 40.1%, H/L: 40.3%, NHW: 34.1%) but there was little evidence that this unadjusted difference was significant between the groups (p = 0.40). Additionally, no significant unadjusted differences were found between the groups with respect to having ever been administered epinephrine. NHB and H/L children less frequently reported having ever used epinephrine compared to NHW children (NHB: 30.1%, H/L: 26.0%, NHW: 36.0%)

Table 1.

Demographic and Clinical Characteristics of Cohort by Race and Ethnicity

	Total N=784	NHW N=425	NHB N=282	H/L N=77	p-value
Sex					0.57
Male	484 (61.7%)	257 (60.5%)	181 (64.2%)	46 (59.7%)
Female	300 (38.3%)	168 (39.5%)	101 (35.8%)	31 (40.3%)
Institution					<0.001
Northwestern/Lurie	277 (35.3%)	151 (35.5%)	80 (28.4%)	46 (59.7%)
Cincinnati Children's	186 (23.7%)	122 (28.7%)	60 (21.3%)	4 (5.2%)
Children's National	175 (22.3%)	98 (23.1%)	69 (24.5%)	8 (10.4%)
Rush Children's	146 (18.6%)	54 (12.7%)	73 (25.9%)	19 (24.7%)
Age at Enrollment	5.9 (3.6)	5.3 (3.5)	7.0 (3.5)	5.9 (3.8)	<0.001
Age at First Reaction	1.4 (1.7)	1.1 (1.4)	1.7 (2.1)	1.4 (1.5)	<0.001
Number of Distinct FA	3.3 (1.8)	3.2 (1.7)	3.6 (2.0)	2.9 (1.7)	0.009
Maximum Reaction Severity					0.40
Mild	107 (13.6%)	65 (15.3%)	30 (10.6%)	12 (15.6%)
Moderate	266 (33.9%)	153 (36.0%)	91 (32.3%)	22 (28.6%)
Severe	289 (36.9%)	145 (34.1%)	113 (40.1%)	31 (40.3%)
Life in Danger	89 (11.4%)	50 (11.8%)	31 (11.0%)	8 (10.4%)
Missing	33 (4.2%)	12 (2.8%)	17 (6.0%)	4 (5.2%)
Ever Administered
Epinephrine					0.14
No	508 (64.8%)	266 (62.6%)	187 (66.3%)	55 (71.4%)
Yes	258 (32.9%)	153 (36.0%)	85 (30.1%)	20 (26.0%)
Missing	18 (2.3%)	6 (1.4%)	10 (3.5%)	2 (2.6%)

Table 2 describes the results from the mixed-effects ordinal logistic regression model of reaction severity. It was determined that NHB children had 70% increased odds of a 1-level increase in reaction severity (OR: 1.7 (95% CI: 1.2, 2.3). Notably, age at first reaction, sex, age at enrollment, and enrollment site did not significantly affect reaction severity. In Table 3, NHB children had a significantly higher odds of going to the ER compared to NHW children (OR: 2.8, 95% CI: 1.4, 5.4) even after controlling for reaction severity and receiving epinephrine. However, H/L children had similar odds of visiting the ER compared to NHW children (OR: 1.3, 95% CI: 0.4, 4.0). We found that receiving epinephrine was also associated with an increase in the odds of utilizing the ER (OR: 7.9, 95% CI: 4.1, 15.0) and an increase in reaction severity was strongly associated with an increase in ER utilization for a given reaction (OR: 11.6, 95% CI: 7.3, 18.4). Age at first reaction, age at enrollment, and sex did not significantly affect the odds of ER usage for a given reaction.

Table 2.

Mixed-Effects Ordinal Logistic Regression Model of Reaction Severity

Race
Ref: NHW	Ref
NHB	1.7*** (1.2,2.3)
H/L	1.1 (0.7,1.8)
Age at First Reaction	1.0 (0.9,1.1)
Age at Enrollment	1.0 (1.0,1.1)
Sex
Ref: Male	Ref
Female	0.9 (0.7,1.2)
Site
Ref: Northwestern/Lurie	Ref
Cincinnati Children's	0.7 (0.5,1.0)
Children's National	1.0 (0.7,1.4)
Rush Children's	1.0 (0.7,1.4)
Observations	1,867 reactions 737 participants

Table 3.

Mixed-Effects Logistic Regression Model of ER Usage

Race
Ref: NHW	Ref
NHB	2.8** (1.4, 5.4)
H/L	1.3 (0.4, 4.0)
Age at First Reaction	0.9 (0.7,1.1)
Age at Enrollment	1.0 (0.9, 1.0)
Sex
Ref: Male	Ref
Female	0.7 (0.4,1.3)
Received Epinephrine
Ref: No	Ref
Yes	7.9*** (4.1, 15.0)
Reaction Severity Site	11.6*** (7.3, 18.4)
Ref: Northwestern/Lurie	Ref
Cincinnati Children's	0.6 (0.3,1.4)
Children's National	0.4 (0.2,1.0)
Rush Children's	0.5 (0.2,1.1)
Observations	1,808 reactions 732 participants

Table 4 describes the results of the Poisson regression of total lifetime epinephrine use. The incidence of total lifetime epinephrine use was 60% less for NHB children (IRR: 0.4, 95% CI: 0.3, 0.5) and 70% less for Hispanic/Latino (IRR: 0.3, 95% CI: 0.2, 0.5) children compared to NHW. Further, an increase in maximum reaction severity was associated with a 40% increase in the incidence of total lifetime epinephrine (IRR: 1.4, 95% CI: 1.2, 1.6). We also found that as the age at first reaction increased by 1 year, the incidence of lifetime epinephrine use decreased by approximately 20% (IRR: 0.8, 95% CI: 0.7, 0.9). Having ever utilized the ER during a food allergy reaction was also associated with an increase in total lifetime epinephrine usage (IRR: 2.0, 95% CI: 1.6, 2.5). Finally, compared to Lurie Children’s Hospital, we found that children enrolling from Cincinnati Children’s Hospital and Rush Children’s Hospital had a decreased incidence of lifetime epinephrine use: (Cincinnati IRR: 0.5, 95% CI: 0.4, 0.7 and Rush IRR: 0.4, 95% CI: 0.3, 0.5).

Table 4.

Poisson Regression Model of Total Lifetime Epinephrine Use

Race
Ref: NHW	Ref
NHB	0.4*** (0.3,0.5)
H/L	0.3*** (0.2,0.5)
Age at First Reaction	0.8*** (0.7, 0.9)
Age at Enrollment	Offset
Sex
Ref: Male	Ref
Female	1.1 (0.9,1.4)
Ever Utilized ER
Ref: No	Ref
Yes	2.0*** (1.6, 2.5)
Maximum Reaction Severity Site	1.4*** (1.2, 1.6)
Ref: Northwestern/Lurie	Ref
Cincinnati Children's	0.5*** (0.4,0.7)
Children's National	1.1 (0.9,1.4)
Rush Children's	0.4*** (0.3,0.5)
Observations	719 participants

Discussion

In this prospective, multi-site cohort study of a diverse population of children with FA, we found that NHB children had a higher odds of reaction severity compared to NHW children, a higher odds of ER usage, and a lower rate of total lifetime epinephrine use compared to NHW children. While we did not find significant differences between H/L and NHW participants with respect to reaction severity and ER usage, we did find that H/L participants had a lower incidence of total lifetime epinephrine use. To our knowledge, these findings represent the first characterization of both FA reaction severity and healthcare utilization differences by race and ethnicity. These findings suggest that when NHB children have a severe reaction, they are more likely to be taken to the ER and less likely to receive epinephrine.

We suspect that this observation is largely driven by gaps in understanding food allergy and reaction management, as we have shown in a prior FORWARD cohort study that NHB participants had lower overall FA knowledge scores when compared to NHW participants.⁹ This knowledge gap may provide some insight into our findings on reaction severity and EAI usage. In this prior study, higher FA knowledge scores were also consistently associated with lower odds of multiple potentially risky FA purchasing and eating behaviors.⁹ Discrepancies in these behaviors may contribute to the disparities seen in ER usage rates, as avoidance of risky FA purchasing and eating behaviors can minimize the risk of accidental allergen exposure and resultant ER utilization. Parental knowledge may also impact their child’s treatment course following an accidental allergen exposure. Caregivers with more knowledge about FA may be more likely to recognize and treat mild or atypical reactions that may not be easy to attribute to FA. Additionally, if a child is presenting with symptoms of severe food allergy reaction and parents/caregivers are unsure of what to do, parents may bring their child to the ER without providing immediate and effective at-home interventions. Short durations between allergen exposure and treatment administration are protective against progression to more severe reactions, and parental food allergy knowledge deficits can lead to delays in recognizing food allergy symptoms and intervening quickly following exposures.⁵

In addition to education, we suspect that socioeconomic differences in this cohort may have also contributed to these disparities. We previously reported that among the FORWARD cohort, NHB participants were more frequently lower-income compared to NHW participants. Given this, at least part of the more than 50% reduction in lifetime epinephrine use may be attributable to a lack of access to epinephrine auto-injectors. While we have demonstrated that the out-of-pocket costs of EAIs does not significantly differ by race and ethnicity, primary access to EAI may still be inadequate.¹¹ It is possible that resource-limited families may struggle to pick up prescriptions due to time constraints or lack of reliable transportation, and ongoing pharmacy shortages of EAI stock may disproportionately affect resource-limited areas. This is further supported by the differences in lifetime EAI usage by clinical sites – two clinical sites: participants from Rush Children’s Hospital and Cincinnati Children’s Hospital were observed to have a significantly lower lifetime EAI usage compared to those children enrolled from Lurie Children’s Hospital. Given that Rush and Cincinnati Children’s Hospital enrolled FORWARD participants from particularly resource-limited neighborhoods in their respective cities, this observation further supports a socioeconomic and access disparity mechanism for these findings.

Finally, we found that as the age at first reaction increased, total lifetime EAI usage decreased, despite controlling for the length of exposure to severe reactions and subsequent epinephrine use through the use of an offset for age of the child. We hypothesize that, as the age at first reaction is delayed, children may be more equipped with the tools needed to avoid food-allergy risk, such as health education and self-efficacy, thus reducing the risk of reactions requiring epinephrine use.

There are several important limitations of this study to consider. First, the reactions that were analyzed in the study are necessarily a more severe subset of all reactions experienced by the participant. This is due to the nature of the survey which first asked participants to recall whether there as a history of any reactions for a given allergen, and subsequently to describe the details (symptoms, severity, and management decisions) of the most severe reaction for that given allergen. Given that we do not capture information on less severe reactions to each allergen, we are limited in the generalization of these conclusions to all allergic reactions, both severe and non-severe. Another limitation we encountered is that the survey administered did not distinguish between pre-hospital EAI and hospital-administered EAI. In this study, we report total EAI usage but given the lack of granularity, we cannot comment on whether these results apply to either pre-hospital or in-hospital EAI alone. Finally, though our model inferences were robust to various specifications, model misspecification, residual confounding, and unmeasured confounding all may threaten the validity of these conclusions.

We conclude that patient race is associated with differences in FA management, and our findings suggest that FA management is suboptimal in NHB children when compared to their NHW peers. This result provides important context that pediatricians must consider when caring for patients with FA of all backgrounds. Further research is needed to better the understand specific mechanisms by which racial and socioeconomic factors impact FA outcomes, with the goal of informing future changes to clinical practice and advocacy efforts to promote equality in FA outcomes.

Highlights:

1. What is already known about this topic?

Food allergy reactions, especially those that progress to anaphylaxis, are hypothesized to be affected by socioeconomic factors such as income and race and ethnicity but little data exists to describe the relationship between these factors.

2. What does this article add to our knowledge?

We demonstrate that non-Hispanic Black children had higher odds of severe reactions and increased ER usage, and both non-Hispanic Black and Hispanic/Latino children had lower rates of lifetime epinephrine use compared to non-Hispanic White children.

3. How does this study impact current management guidelines?

Patients who report severe reactions and demonstrate increased ER utilization with decreased epinephrine use should be offered resources and training on definitive management of various food allergy reactions.

List of Abbreviations

NHW

non-Hispanic White

NHB

non-Hispanic Black

H/L

Hispanic/Latino

ER

Emergency Room

EAI

Epinephrine Auto-Injector

US

United States

FORWARD

Food Allergy Outcomes Related to White and African American Racial Differences

FA

Food Allergy

ANOVA

Analysis of Variance

OR

Odds Ratio

CI

Confidence Interval

IRR

Incidence Rate Ratio