Early-life environmental exposures associate with individual and cumulative allergic morbidity

Abstract

Several early-life environmental factors have been associated with altered risk for the development and/or severity of individual allergic conditions. These include exposures implicated in the modulation of the microbiome, such as infant delivery mode, diet, and exposure to antibiotics and antacids. The impact of these early-life factors on allergic multimorbidity remains unknown. To address this knowledge gap, we used electronic medical records for a birth cohort of 158,510 children to track development of atopic dermatitis (AD), IgE-mediated food allergy (IgE-FA), asthma, and allergic rhinitis (AR) in individual children over time. We measured hazard ratios (HRs), adjusted for birth year, race, ethnicity, sex, and insurance payer type, to assess how development of both individual and multiple allergic conditions is influenced by birth mode, feeding practice during the first year of life, or exposure to antibiotics and/or antacids during the first six months of life. We found that vaginal delivery (VD; HR 0.89, 0.83, 0.84, 0.79 for at least 1, 2, 3, 4 conditions, respectively; p≤0.001) and exclusive breastmilk (BM) feeding (HR 0.74, 0.75, 0.89, for at least 1, 2, 3 conditions, respectively; p≤0.001) are associated with reduced cumulative allergic burden, while antibiotic exposure (HR 1.40, 1.44, 1.48, 1.63 for at least 1, 2, 3, 4 conditions, respectively; p≤0.001) and antacid exposure (HR 1.26, 1.35, 1.32 for at least 1, 2, 3 conditions, respectively; p≤0.001) are associated with increased cumulative allergic burden during childhood. This work expands our understanding of how a child’s early-life environment may influence their risk of allergy development and progression.

To the Editor,

Factors implicated in aberrant microbial colonization, or dysbiosis, during infancy may contribute to the development of allergic disease.¹ Among them are birth by cesarean section (CS) rather than by vaginal delivery (VD), infant consumption of formula rather than breastmilk (BM), and early-life exposure to antibiotics or antacids.^2–7 Prior studies examining the influence of these early-life environmental factors on allergic outcomes have largely focused on individual allergic conditions and have not considered their impact on cumulative allergic morbidity. Knowing that a substantial portion of pediatric patients ultimately suffer from more than one allergic manifestation,⁸ we studied whether birth delivery mode, infant feeding practice, and/or early-life exposure to antibiotics or antacids are associated with cumulative allergic disease burden during childhood.

Using an electronic medical record (EMR)-based (EPIC Systems, Verona, WI) primary care cohort comprised of 158,510 children seen at Children’s Hospital of Philadelphia, we identified subjects with atopic dermatitis (AD), IgE-mediated food allergy (IgE-FA), asthma, and allergic rhinitis (AR) based on presence of appropriate ICD codes, documentation, and medication prescriptions, as previously described and further detailed in Supporting Information.^8,9 We also defined subjects with 0 and at least 1, 2, 3, or 4 allergic manifestations. As detailed in Supporting Information, delivery mode and feeding practices during the first year of life were queried using specified search terms, and EMR-documented prescription of antibiotics or antacids during the first six months of life was used as a surrogate for exposure to medications in these classes.

Using R software (version 3.6.1), we performed retrospective case-control comparisons using Cox proportional hazards monitoring with adjustments for demographic covariates, namely birth year, race, ethnicity, sex, and insurance payer type. First, we examined the rate of onset of individual allergic diseases in the exposure and control cohorts. Next, we examined the onset of multiple allergic diseases in the exposure and control cohorts. Because of variable follow-up time, we chose Cox proportional hazard ratios with confidence intervals to present our results. Kaplan-Meier curves were plotted to display outcomes over time. After adjustment for covariates, p values less than .05 were considered significant. We observed comparable HRs between our analyses of single environmental factors and subsequent analysis in which all exposures were considered simultaneously (data not shown). For each allergic condition, charts from 100 randomly selected patients were manually reviewed for presence of allergic conditions and environmental exposures and showed concordance with our bioinformatic analysis. Our study was reviewed by the CHOP Institutional Review Board and did not meet the definition of human subjects research.

Subject demographic features are detailed in Table 1 based on allergic diseases and in Supporting Table 1 based on exposures. Additionally, numbers of subjects with 1–4 allergic conditions and associated exposures are listed in Supporting Table 2. Sex distribution was approximately equal in the full primary care birth cohort (51% males vs. 49% females). There was a significantly increased prevalence of males and black individuals in the allergic groups as compared to the overall birth cohort (57% males vs. 51% overall and 45% blacks vs. 32% overall; p≤0.001 for both comparisons by chi-square testing).

Table 1.

Demographic characteristics of full and allergic disease-specific cohorts

	Full (158,510)	AD (18,596)	IgE-FA (6,472)	Asthma (29,351)	AR (27,094)
Sex, % (n)
Male	51.3 (81,266)	54.9 (10,210)	59.0 (3,819)	59.7 (17,512)	56.1 (15,200)
Female	48.7 (77,244)	45.1 (8,386)	41.0 (2,653)	40.3 (11,839)	43.9 (11,894)
Race, % (n)
White	49.6 (78,691)	32.1 (5,964)	46.9 (3,034)	39.8 (11,688)	43.2 (11,698)
Black	32.1 (50,847)	50.9 (9,465)	34.4 (2,227)	46.0 (13,504)	43.8 (11,860)
Asian or Pacific Islander	3.7 (5,824)	5.1 (957)	6.9 (444)	2.7 (785)	2.8 (748)
Other	2.0 (3,158)	1.7 (312)	2.2 (140)	1.9 (548)	1.5 (416)
Unknown	12.6 (19,990)	10.2 (1,898)	9.7 (627)	9.6 (2,826)	8.8 (2,372)
Ethnicity, % (n)
Hispanic or Latino	6.6 (10,495)	5.9 (1,094)	4.4 (283)	6.7 (1,957)	5.8 (1,578)
Non-Hispanic or Latino	93.4 (148,015)	94.1 (17,502)	95.6 (6,189)	93.3 (27,394)	94.2 (25,516)
Birth year, % (n)
2000 to 2004	8.4 (13,357)	13.5 (2,512)	9.3 (602)	12.7 (3,716)	15.4 (4,173)
2005 to 2009	38.2 (60,576)	42.1 (7,829)	45.0 (2,912)	47.4 (13,915)	53.4 (14,468)
2010 to 2014	39.2 (62,076)	31.8 (5,919)	38.0 (2,458)	34.9 (10,246)	28.8 (7,793)
2015 or later	14.2 (22,501)	12.6 (2,336)	7.7 (500)	5.0 (1,478)	2.4 (660)
Payer type, % (n)
Medicaid	33.0 (52,289)	43.0 (7,994)	25.9 (1,678)	42.1 (12,364)	37.6 (10,175)
Non-Medicaid	67.0 (106,221)	57.0 (10,602)	74.1 (4,794)	57.9 (16,987)	62.4 (16,919)

AD, atopic dermatitis; IgE-FA, IgE-mediated food allergy; AR, allergic rhinitis

We first examined the effects of early-life environmental factors on the development of individual allergic conditions (Figure 1). We found that VD, as opposed to CS, was associated with reduced rate of development of IgE-FA (HR 0.86, p≤0.001), asthma (HR 0.85, p≤0.001), and AR (HR 0.84, p≤0.001); there was no association between birth mode and development rate of AD (Figure 1A). Compared to formula, BM was associated with reduced rate of developing AD (HR 0.91, p≤0.001), asthma (HR 0.58, p≤0.001), and AR (HR 0.73, p≤0.001) (Figure 1B). This relationship was also apparent for formula-supplemented BM and development of asthma (HR 0.85, p≤0.001) and AR (HR 0.95, p≤0.001). Notably, exclusive BM and formula-supplemented BM were associated with increased rate of developing IgE-FA (HR 1.58 and HR 1.37, respectively; p≤0.001). However, in additional analyses in which IgE-FA diagnosis was limited to age ≤5, exclusive BM was associated with significantly decreased rate of developing IgE-FA (HR 0.82; p≤0.001) (Supporting Table 3). Lastly, we observed increased rate of developing AD, IgE-FA, asthma, and AR in association with early-life exposure to antibiotics (HR 1.34, HR 1.32, HR 1.53, HR 1.23; p≤0.001) and antacids (HR 1.09, HR 1.11, HR 1.37, and HR 1.31; p≤0.001, except p≤0.01 for IgE-FA) (Figure 1C,D).

Figure 1.

Kaplan-Meier curves demonstrating associations between (A) delivery mode, (B) feeding mode, and early-life exposure to (C) antibiotics and (D) antacids and the rate of developing atopic dermatitis, IgE-mediated food allergy, asthma, or allergic rhinitis. Dashed lines represent 95% confidence intervals. *, reference group; HR, hazard ratio; NS, not significant.

We next examined the effects of early-life environmental factors on allergic multimorbidity (Figure 2). VD was associated with reduced allergy development rate (HR 0.89, 0.83, 0.84, 0.79 for at least 1, 2, 3, 4 conditions, respectively; p≤0.001) (Figure 2A). Exclusive BM (HR 0.74, 0.75, 0.89, for at least 1, 2, 3 conditions, respectively; p≤0.001) and, to a lesser extent, formula-supplemented BM (HR 0.94 for at least 1 condition; p≤0.001) were likewise associated with reduced allergy development rate (Figure 2B). Lastly, early-life antibiotic exposure (HR 1.40, 1.44, 1.48, 1.63 for at least 1, 2, 3, 4 conditions, respectively; p≤0.001) and antacid exposure (HR 1.26, 1.35, 1.32 for at least 1, 2, 3 conditions, respectively; p≤0.001) were associated with increased allergy development rate (Figure 2C,D). For both feeding mode and antacid exposure, there were insufficient subjects to detect associations with development of 4 allergic conditions.

Figure 2.

Kaplan-Meier curves demonstrating associations between (A) delivery mode, (B) feeding mode, and early-life exposure to (C) antibiotics and (D) antacids and the rate of developing at least 1, 2, 3, or 4 allergic conditions. Dashed lines represent 95% confidence intervals. *, reference group; HR, hazard ratio; NS, not significant.

In summary, our findings indicate that early-life factors associate with both individual and cumulative allergic morbidity. In an era of globally rising CS rates, our findings support careful review of the indications for CS by an individual’s obstetrician; whether restoration of the vaginal microbiome translates to long-term protection from allergic disease remains unclear.¹⁰ Our analysis identified a protective association between BM and allergic multimorbidity. This effect was maintained between BM and IgE-FA diagnosed at age ≤5 but was reversed for subjects diagnosed with IgE-FA at any point during childhood. This may be due to mechanistic differences between early and late onset IgE-FA. We also note that reverse causality is a relevant confounder of the effects of feeding practices on food allergy development, as highlighted in a recent expert review that did not identify a clear relationship between BM and IgE-FA.⁴ Finally, our findings support the judicious use of antibiotics and antacids in infancy, echoing findings recently reported by Mitre et al.⁷

Limitations of our study include issues inherent to retrospective case-control studies.⁹ Among these are possibilities of inaccurate determination of disease diagnosis and/or duration, medication prescriptions from outside our healthcare system, ascertainment bias, and reverse causality (e.g. prolonged breastfeeding duration driven by caregiver concern for allergy). Our chart review helped to negate some of these potential biases. Our regressions accounted for birth year, race, ethnicity, sex, and insurance payer type. We acknowledge that our analysis could not account for other possible confounders, such as history of atopic disease or socioeconomic status.

Our work is consistent with the hypothesis that early-life environmental factors that impact microbial colonization may influence cumulative allergic risk. Mechanistically, dysbiosis may facilitate allergic pathogenesis through blunting of anti-inflammatory responses, causing heightened sensitization to allergens.^1–3,6 Future research on the complex interplay between environmental factors and allergic predisposition will undoubtedly benefit from prospective longitudinal studies, investigation of geographically broader datasets, and consideration of additional environmental factors, including timing of food introduction and nutritional status beyond infancy.

Impact Statement:

Vaginal delivery, breastmilk-based infant diet, and avoidance of antibiotics or antacids during infancy may provide protection against allergic morbidity and multimorbidity during childhood.

Abbreviations:

VD

vaginal delivery

CS

cesarean section

BM

breastmilk

AD

atopic dermatitis

IgE-FA

IgE-mediated food allergy

AR

allergic rhinitis

HR

hazard ratio