COVID-19 and Asthma Onset in Children

Data were abstracted from the electronic health records (EHR) of children ages 1 to 16 who received care within the Children’s Hospital of Philadelphia (CHOP) Care Network from 2019 to 2022. We limited our analysis to subjects who obtained polymerase chain reaction (PCR) testing for SARS-CoV-2 between March 1, 2020 and February 28, 2021. Of note, SARS-CoV-2 PCR testing became available at CHOP in March 2020, when the first incident case was reported.¹⁰  The interval between March 1, 2020 and February 28, 2021 preceded both the introduction of routine rapid antigen testing and the introduction of vaccination for pediatric patients. All subjects were considered unvaccinated during the exposure window.

All subjects were required to have at least 1 ambulatory well child visit in the CHOP Care Network (CCN) in the year preceding the PCR test and at least 1 CCN visit at any time during an 18-month follow-up period after their first positive or last negative PCR test (if all test results were negative). A CCN follow-up visit could include primary care, urgent care, emergency department visits, telemedicine, or inpatient admission. Additionally, this 18-month follow-up period began 30 days after the corresponding PCR test. A 30-day lag time between the exposure and the start of the observation period was chosen to provide a reasonable separation between measurement of the exposure and the outcome. This excluded study subjects who may have received an asthma diagnosis on the same day or very close to the day of their SARS-CoV-2 PCR test.

Our primary definition of asthma diagnosis was the presence of at least 1 corresponding International Classification of Diseases (ICD) code for asthma (codes 493.×, J45.×) at the CCN visit, along with a prescription of an asthma-related medication. ICD codes related to “reactive airway disease” and “postviral wheeze” were excluded. An asthma-related medication included any of the following: β agonists, inhaled corticosteroids, combined inhaled corticosteroids and long-acting β agonists, leukotriene modulators, or biologics. For a sensitivity analysis, a secondary definition of asthma diagnosis was used that required a corresponding ICD code for asthma at 2 or more follow-up visits at least 6 months apart, as well as prescription of an asthma-related medication.

Sex, age, race, and insurance status of study subjects were included for analysis. Although we acknowledge that race and ethnicity are social constructs, race was included as a variable in our analyses given prior research showing that Black and Latinx children have higher asthma prevalence and greater asthma morbidity, most likely because of a combination of factors, including structural racism, healthcare inequities, differential environmental exposures, and the effects of chronic, long-term stress.¹¹  The Childhood Opportunity Index is a validated approach for estimating social determinants of health across various neighborhood-level indices,¹²  and this was used in our study to approximate socioeconomic status, housing quality, and the built environment. Gestational age, BMI, and atopic comorbidities (specifically atopic dermatitis, food allergies, and allergic rhinitis), were also included as potential confounders.

The covariates of interest were summarized and compared between SARS-CoV-2 positive and SARS-CoV-2 negative groups using Pearson’s χ-square test for categorical variables and 2-sample t test for continuous variables (Table 1). Multivariable Cox regression models assessed the hazard of new asthma diagnosis comparing SARS-CoV-2 positive and SARS-CoV-2 negative groups. Models were adjusted for the covariates outlined in Table 1. Subgroup analyses were performed based on the age groups defined by the National Heart, Lung, and Blood Institute guidelines (ages 0–4, 5–11, and ≥12 years) given the phenotypic similarities within each group.¹³  The subgroup analyses used the primary definition of asthma for the outcome variable.

All analyses were completed using Stata 15.1 (Stata Corp, College Station, TX). A significance level of P < .05 was chosen a priori for all analyses. The CHOP Institutional Review Board determined this study was not human subject’s research.

A description of the study sample (n = 27 423) appears in Table 1. Of the subjects that met inclusion criteria, 3147 subjects (11.5%) were in the SARS-CoV-2 positive group and 24 276 (88.5%) were in the SARS-CoV-2 negative group. SARS-CoV-2 positive individuals tended to be older, Black, Medicaid insured, among lower childhood opportunity index quintiles, and have a higher BMI (Table 1). Regarding atopic comorbidities, SARS-CoV-2 positive individuals were less likely to have a food allergy diagnosis but more likely to have an allergic rhinitis diagnosis than SARS-CoV-2 negative individuals (Table 1).

There were 573 total subjects who received an asthma diagnosis in the 18-month follow-up period. In our primary analysis of these subjects, 1.81% (n = 57) of SARS-CoV-2 positive subjects were subsequently diagnosed with asthma compared with 2.13% (n = 516) of SARS-CoV-2 negative subjects. Overall, SARS-CoV-2 positive subjects trended toward lower incidence of new asthma diagnosis at most time points over the 18-month follow-up period (Fig 1). In the regression analyses (Table 2), SARS-CoV-2 PCR positivity was not associated with the hazard of new asthma diagnosis in the 18-month follow-up period (hazard ratio [HR]: 0.96; 95% confidence interval [CI]: 0.73–1.27). Black race (HR: 1.49; 95% CI: 1.13–1.95), comorbid food allergy (HR: 1.26; 95% CI: 1.03–1.55), and allergic rhinitis (HR: 2.30; 95% CI: 1.93–2.74) were associated with a significantly higher hazard of new asthma diagnosis, whereas older age (ages 5–11: HR: 0.27; 95% CI: 0.22–0.34; ages 12+: HR: 0.16; 95% CI: 0.12–0.22) was associated with a significantly lower hazard of new asthma diagnosis (Table 2).

In our sensitivity analysis utilizing our secondary asthma definition that required an asthma ICD code at 2 or more follow-up visits, there were 418 total subjects who received and maintained an asthma diagnosis in the 18-month follow-up period. Of those, 1.12% (n = 35) of SARS-CoV-2 positive subjects were diagnosed with asthma compared with 1.59% (n = 383) of SARS-CoV-2 negative subjects. In the regression analyses (Table 3), SARS-CoV-2 PCR positivity was not associated with a new asthma diagnosis using this stricter definition. However, Black race, atopic comorbidities, and age were again associated with new asthma diagnosis when using the secondary asthma definition (Table 3).

For each subgroup analysis by age, there was no significant relationship between SARS-CoV-2 positivity and new asthma diagnosis in each of the 3 age categories (Tables 4–6). Among toddlers, race and atopic comorbidities were again associated with a higher hazard of new asthma diagnosis in the subsequent 18 months. Additionally, toddlers who were born preterm or had a higher BMI also had a significantly higher hazard of new asthma diagnosis in the subsequent 18 months (Table 4). Female sex significantly increased the hazard of new asthma diagnosis among school-age children and teenagers (Tables 5 and 6). All other risk factors (except for allergic rhinitis in school-age children) had no significant relationship with new asthma diagnosis among school-age children and teenagers (Tables 5 and 6).

In this longitudinal cohort study of children observed 1 year before and 18-months after receiving SARS-CoV-2 PCR testing in the first year of the pandemic, we found that SARS-CoV-2 PCR positivity was not associated with subsequent asthma diagnosis. Although contrary to the literature regarding other respiratory viral infections, particularly RSV and rhinovirus, our findings are consistent with contemporary SARS-CoV-2 literature. For example, recent descriptive epidemiology studies have demonstrated a similar incidence of asthma diagnoses between pre and postpandemic levels, despite widespread transmission of SARS-CoV-2.^14 ,15  Additionally, the Long COVID Research Group demonstrated that long-term pulmonary function is not affected in both symptomatic and asymptomatic children infected with the SARS-CoV-2 virus.¹⁶ 

Defining asthma phenotypes in a pediatric cohort is challenging, especially among younger children who may present with a self-limited acute wheezing episode. Our investigation found a higher incidence of asthma among young children, with older age being associated with a significantly lower likelihood of new asthma diagnosis. It is worth noting that prior epidemiologic analyses of our cohort have demonstrated a high cumulative incidence of asthma (15.9%) in the 0 to 5 age group.¹⁷  This value approximates that which was found in a recent multistate analysis (17.5%) of the 0 to 5 age group,¹⁸  indicating that asthma diagnosis rates in our region are similar to those observed nationally. To minimize the likelihood of transient wheeze confounding our current study, we excluded ICD codes relating to reactive airway and postviral wheeze from our asthma definition. Further, we performed a sensitivity analysis that required more than 1 encounter with an asthma diagnosis. The results of this analysis did not differ significantly from our primary analysis, indicating that transient wheeze episodes likely did not substantially alter our findings.

The lack of observed association between SARS-CoV-2 and incident asthma is also supported by the consistency of the association between other asthma risk factors and new asthma diagnosis in the current study. In the primary analysis, atopic history (specifically food allergies and allergic rhinitis) was associated with a significantly higher risk of asthma diagnosis. Atopy is a well-studied risk factor for the development of asthma given its biologic predisposition to other allergic conditions.¹⁹  As demonstrated by our subgroup analyses, these relationships are especially relevant for infants and toddlers when lung development is most vulnerable to complex differences in both biologic and environmental factors. Notably, the subgroup analysis of toddlers yielded significant relationships between preterm gestation and high BMI, both of which increased the risk of developing asthma. These are also well-studied risk factors across the literature, particularly among young children.^20 ,21 

This study further benefits from a relatively unique period in history in which there was more uniformity in and control over environmental risk factors for asthma development. Principally, the prevalence of other respiratory viral illnesses (including RSV, influenza, and rhinovirus) plummeted during the first year of the pandemic.¹⁰  This fact provided a natural opportunity to isolate the effects of SARS-CoV-2 from other respiratory viral infections. Additionally, global improvement in pollution trends and air quality, coupled with greater isolation indoors, may have helped minimize exposure to outdoor toxicants and allergens while also minimizing variability between urban and rural settings.²²  Further, the pandemic generated a widespread practice of PCR testing of both symptomatic and asymptomatic individuals (for example, in the cases of preprocedural and preadmission testing). Typically, asymptomatic children are not tested for viral infections, making it difficult to define an epidemiologic comparison group that lacks the exposure of interest. In other words, the fact that healthcare systems were performing the gold-standard test on asymptomatic children during the first year of the pandemic created a natural experimental design that facilitated the current study.

Recent investigation has shed light on possible pathophysiologic mechanisms that may better inform the understanding of our results. SARS-CoV-2 infection tends to elicit an immunophenotype notable for a robust type I inflammatory response characterized by type I interferon and inflammasome activation and subsequent CD8 T cell response.²³  This is well demonstrated in the context of the multisystem inflammatory syndrome in children, an immunopathologic response characterized by overactivation of these pathways.²³  Infection with the SARS-CoV-2 virus has also been shown to induce eosinopenia regardless of disease severity.^24 ,25  This is contrary to infection with RSV or rhinovirus, which have been linked to increases in certain atopic biomarkers, such as peripheral eosinophilia.¹  Taken altogether, infection with the SARS-CoV-2 virus may drive the cellular immunophenotype at the level of the airway epithelium to a more type I, as opposed to type II, inflammatory state.^26 ,27  Chronic asthma, on the other hand, is primarily mediated by type II inflammation characterized by IL-4, IL-5, IL-13, and eosinophilia.^27 ,28 

There are inherent limitations to our study. First, the use of EHR data from 1 institution may impact the generalizability of our findings. However, the CCN covers a 3-state region in one of the most populated parts of the country, including urban and suburban settings. Further, the CCN patient population is largely reflective of allergic disease patterns across the country, at least in the younger age group.¹⁸  Nonetheless, EHR data are also limited in its reliance upon provider-entered information. For example, our EHR data does not consistently capture certain risk factors, such as family history of asthma and secondhand smoke exposure. In addition, our study relies upon SARS-CoV-2 PCR test results for a 1-year exposure window, which may have misclassified individuals who later obtained a positive PCR or rapid antigen test after the exposure window. However, this misclassification is unlikely to change the direction of the association given the consistent relationship between positive and negative SARS-CoV-2 exposure groups over time in our analyses. Studying the first year of the pandemic was an important feature of our experimental design as it isolated a time interval before the deployment of home rapid antigen SARS-CoV-2 testing. However, patients seeking care during this period may have been less able to delay care until pandemic conditions improved. As such, our study population may have had a higher medical complexity than the US pediatric population as a whole. Because of the limited number of children with severe primary SARS-CoV-2 infection, low subject numbers precluded considering SARS-CoV-2 severity in our analyses. Therefore, future studies may identify subcohorts of subjects where SARS-CoV-2 infection may modify incident asthma risk. Finally, as our exposure window preceded the evolution of several SARS-CoV-2 variants (eg, ο), it is possible that these later SARS-CoV-2 strains may influence asthma risk differently.

Taken altogether, we find a consistent finding across all pediatric age groups that SARS-CoV-2 PCR positivity does not confer an additional risk for asthma diagnosis, at least within the first 18 months after PCR test. This information may be useful for families and providers alike in the prognostication of the long-term respiratory effects after SARS-CoV-2 infection in children. As one of the first population-based studies to investigate SARS-CoV-2 infection in asthma-naïve children, this study has several implications. This study refines our knowledge regarding the long-term respiratory effects of SARS-CoV-2 infection in children and adds to our growing understanding of how SARS-CoV-2 may influence asthma development in pediatric patients. Future work should consider SARS-CoV-2 infection severity and stratify between asymptomatic and symptomatic children and between those with mild, moderate, or severe symptoms. Future work should also consider asthma by the persistence and severity of symptoms in children with both new and prior asthma diagnoses.