Asthmalitis? Diagnostic Variability of Asthma and Bronchiolitis in Children <24 Months Free

This was a retrospective cross-sectional study using the Pediatric Health Information System (PHIS) database. PHIS includes clinical and billing data from tertiary pediatric hospitals affiliated with the Children’s Hospital Association (Lenexa, KS) in the United States.¹³  Data quality is ensured by participating hospitals in conjunction with the Children’s Hospital Association. The PHIS data were accessed as a limited data set, with each patient having a unique encrypted identifier so repeat encounters can be tracked longitudinally. We included the 42 hospitals with complete clinical and billing data for emergency department and inpatient encounters during the study period. This study was approved by the institutional review board at our institution.

We included children <2 years of age with an ED or inpatient hospital encounter between January 1, 2017 and December 31, 2021 with a primary or secondary International Classification of Diseases, Tenth Revision (ICD-10) diagnostic code for asthma (J45), bronchiolitis (J21), wheeze (R06.2), or bronchospasm (J98.01). We chose these diagnoses with the goal of describing and comparing children with bronchiolitis and asthma while not missing those in whom the diagnosis was uncertain or for whom clinicians were hesitant to make a chronic diagnosis of asthma. Although some of these diagnoses, such as reactive airway disease, map to asthma diagnosis codes, wheeze, and bronchospasm, are separate diagnoses and were included for completeness. We used both primary and secondary diagnoses because diagnoses such as acute respiratory failure are commonly the primary diagnosis for children with bronchiolitis.¹⁴  We excluded children with complex chronic conditions using the corresponding flag in PHIS¹⁵  because clinicians likely approach the diagnosis and treatment of these children differently. We excluded children with croup because steroid use was one of our outcomes.

The primary outcome was the diagnosis assigned. We categorized encounters into 4 groups on the basis of the inclusion diagnoses: asthma (without bronchiolitis), bronchiolitis (without asthma), both asthma and bronchiolitis diagnoses, or only a diagnosis of wheeze or bronchospasm. Secondary outcomes included the use of inhaled SABA (defined as a charge for albuterol, levalbuterol, or albuterol with ipratropium) and the use of systemic corticosteroids (defined as a charge for enteral or parenteral dexamethasone, methylprednisolone, prednisolone, or prednisone).

We described demographic characteristics, resource use, and diagnoses assigned using counts and proportions for all categorical variables. In the first part of the analysis, we grouped encounters by month of age of the patient to assess patterns in diagnoses and treatments by age. Month of age was defined as the number of complete months of life from the date of birth. We summarized and plotted the overall number of encounters by month of age, as well as the proportion of encounters with each diagnosis and the proportions of encounters using SABAs or corticosteroids.

In the second part of the analysis, we limited the population to children 12 to 23 months of age to examine factors associated with the diagnoses of interest. The rationale for choosing this age group is based on guideline differences, conflicting or minimal evidence, and symptom overlap that may be more significant than in infants <1 year of age.^2,3  We described the unadjusted proportions of encounters at each individual hospital assigned a diagnosis of asthma, both asthma and bronchiolitis, bronchiolitis, and wheeze or bronchospasm. To assess patient factors associated with different diagnoses, we used univariate and multivariable generalized linear mixed-effects models using random intercepts to adjust for clustering by hospital. The outcome variable was a non-bronchiolitis diagnosis (including diagnosis groups of asthma, asthma with bronchiolitis, and wheeze or bronchospasm). We selected patient factors a priori that have been demonstrated to have a relationship with a diagnosis of asthma, including age, sex,¹⁶  race/ethnicity,¹⁷  history of previous albuterol use (defined as a charge for albuterol at a previous ED or hospital encounter), and number of previous encounters for asthma, bronchiolitis, wheeze, or bronchospasm. We also included level of care (ED, inpatient floor, or ICU, using the highest level of care the patient received in the encounter) both as a marker of illness severity and because clinical approach may differ based on the setting. We excluded encounters with missing sex or race/ethnicity data from the models. As a sensitivity analysis, we performed the same models excluding encounters with only a diagnosis of wheeze or bronchospasm. All statistical analyses were conducted by using R (version 4.2.1; R Foundation for Statistical Computing).

There were 617 862 encounters with a discharge diagnosis of bronchiolitis, asthma, bronchospasm, or wheezing during the 5-year study period. We excluded 57 579 encounters with complex chronic conditions and 6125 encounters with croup, leaving 554 158 encounters for 416 181 unique children in the study. The median age was 8 months, and children 1 to 5 months of age had the largest number of encounters (Fig 1). The cohort was 39% female and 61% male. The majority (75%) of encounters were for bronchiolitis, with 12% having a diagnosis of asthma, 4.3% having a diagnosis of both bronchiolitis and asthma, and 8.5% having only a diagnosis of wheeze or bronchospasm. Across all diagnoses, SABA was used in 35% of encounters, and systemic steroids were used in 20%. The highest level of care was ED for 349 751 (63%) of encounters, inpatient floor for 159 047 (29%), and ICU for 45 360 (8%). Cohort characteristics stratified by age (<12 months vs 12–23 months) are shown in Table 1.

Bronchiolitis was the predominant diagnosis in the youngest infants, making up 98% of encounters <3 months of age. The proportion of the cohort with a bronchiolitis diagnosis decreased with every month of age (Fig 1), making up just 33% of encounters in children at 23 months of age. Asthma diagnoses were uncommon in children <6 months of age but increased in frequency with each month of age to 44% of encounters at 23 months of age. Encounters with only a nonspecific symptom diagnosis of wheeze or bronchospasm were more common in children >12 months of age. More than 50% of encounters included a diagnosis of asthma, wheeze, or bronchospasm starting at age 18 months, and a diagnosis of asthma was more common than a diagnosis of bronchiolitis for children 21 months and older (Supplemental Table 3).

Treatment with SABAs and systemic corticosteroids increased with each month of life. SABA use increased from 8% of encounters in children <1 month of age to 61% of encounters for children aged 23 months. SABAs were used in more than one-half of encounters starting at age 14 months. Corticosteroid use similarly increased with patient age, from 3% in the youngest children to 49% of encounters at 23 months.

Age-related patterns of use of SABAs and steroids varied by diagnosis (Fig 2). For SABAs, children with both asthma and bronchiolitis diagnoses were the highest usage group for all ages. The use of SABAs in children with an asthma diagnosis increased rapidly up to age 4 months and was relatively stable in approximately two-thirds of encounters for all children >6 months of age. Children with bronchospasm or wheeze diagnoses had a more gradual increase in SABA use with age but had greater use than those with asthma in older children. SABA use in children with bronchiolitis increased with each month of age up to 38% to 39% in children 18 months of age and older.

The use of systemic corticosteroids also increased gradually with age. In children with an asthma diagnosis, the use of corticosteroids increased from 9% to 61% over the first year of life and stabilized between 61% to 65% beyond 12 months of age. Children with wheeze or bronchospasm diagnoses followed a similar pattern with slightly lower corticosteroid usage. Children diagnosed with bronchiolitis had considerably lower corticosteroid use at all ages but increased with age from 3% in the youngest children to 19% in the oldest.

Hospitals varied widely in the use of diagnosis codes for asthma, bronchiolitis, bronchospasm, and wheeze. The staff of some hospitals diagnosed up to 5 times more bronchiolitis than asthma, whereas others diagnosed nearly twice as much asthma compared with bronchiolitis in children 12 to 23 months of age. The use of the nonspecific diagnoses of wheeze or bronchospasm (without asthma or bronchiolitis diagnosis assignment) also varied by institution, ranging from 1% to 34% of encounters (Fig 3).

After adjusting for clustering by hospital, there were significant differences in the odds of a non-bronchiolitis diagnosis (asthma, wheeze, or bronchospasm) based on demographic and historical factors in univariate and multivariable mixed-effects models (Table 2). In the multivariable model, the odds of a non-bronchiolitis diagnosis increased with month of age (odds ratio [OR] 1.13, 95% confidence interval [CI] 1.12–1.13), male sex (OR 1.37, 95% CI 1.35–1.40), and non-Hispanic Black race (OR 1.54, 95% CI 1.50–1.58). In comparison with ED discharges, there were lower odds of non-bronchiolitis diagnosis for children treated as inpatients (OR 0.66, 95% CI 0.65–0.68) or in the ICU (OR 0.76, 95% CI 0.73–0.79). The odds of a diagnosis of asthma, wheeze, or bronchospasm increased with the number of previous encounters (OR 2.73, 95% CI 2.61–2.86, for 3 or more encounters), and with history of previous albuterol use (OR 2.24, 95% CI 2.16–2.32).

In the sensitivity analysis, we excluded encounters without an asthma or bronchiolitis diagnosis. Associations between age, sex, race/ethnicity, previous albuterol use, and previous encounters were similar to those found in the primary analysis; however, the relationship was reversed for level of care. Odds of an asthma diagnosis (when excluding encounters for wheeze or bronchospasm) were higher for ICU encounters (OR 1.30, 95% CI 1.25–1.36) than for ED encounters, and not significantly different for inpatient floor encounters (Supplemental Table 4). Because of the contrasting results when excluding encounters with only a diagnosis of wheeze or bronchospasm, we performed a post hoc comparison of the diagnosis group by level of care. Encounters for wheeze or bronchospasm made up 20% of ED encounters in children 12 to 23 months of age but only 3.6% of inpatient encounters and 0.7% of ICU encounters. In contrast, the assignment of both an asthma and a bronchiolitis diagnosis code occurred in 14% of inpatient encounters and 29% of ICU encounters, but only 2.8% of ED discharges (Supplemental Table 5).

This large, multicenter cross-sectional study of acutely wheezing infants and young children in US children’s hospitals revealed that non-bronchiolitis diagnosis codes for asthma, wheeze, or bronchospasm increased steadily with each month of age. Notably, bronchiolitis and non-bronchiolitis diagnoses were approximately equally common in children 12 to 23 months of age, although this varied considerably by institution. The use of SABAs and systemic corticosteroids was more common for children with non-bronchiolitis at all ages.

Guidelines for both bronchiolitis¹  and asthma¹⁸  recommend making the diagnosis clinically. Yet there is considerable clinical overlap in acute presentation, with age often being the discriminating factor.³  Our results call into question the utility of age cutoffs, given the nearly equal frequency of bronchiolitis and non-bronchiolitis diagnosis codes in the 12- to 23-month age group. The wide interhospital variation in these results accentuates this. Some variation is to be expected with differences in patient population, but it is hard to believe that those differences could explain how 1 hospital can diagnose 5 times more bronchiolitis whereas another can diagnose twice as much asthma. Additionally, although some hospitals almost exclusively use diagnosis codes for asthma or bronchiolitis, others use wheeze or bronchospasm in more than one-third of encounters. Part of this variation could simply be due to differences in institutional coding practices because the assignment of asthma diagnoses to children with bronchiolitis may occur if bronchodilators were used during hospitalization.¹⁹  Another potential explanation for these differences is clinical uncertainty, which may be supported by our finding that children hospitalized or in the ICU, in which they may be treated by multiple providers, are more likely to receive diagnosis codes for both asthma and bronchiolitis. There may be appropriate hesitancy to assign a chronic asthma diagnosis during a short ED visit, which could explain the preponderance of nonspecific wheeze or bronchospasm diagnoses in ED encounters.

Provider inexperience with the nuances of respiratory disease in young children could also result in the misdiagnosis of bronchiolitis as asthma. Although only 5.4% of children <12 months of age were diagnosed with asthma in this study of pediatric hospitals, previous work examining diagnostic variation across all care settings in California revealed much higher rates (∼50% of encounters <12 months).²⁰  This stark difference may imply that children treated outside of children’s hospitals may be more likely to receive an inappropriate diagnosis of asthma at young ages. It is likely that misdiagnosis could lead to significant harm; children with true bronchiolitis may be receiving inappropriate and unnecessary treatments whereas those with asthma or recurrent wheeze may not be given needed disease-modifying therapy, such as inhaled corticosteroids.

In this study, groups that are known to have a higher risk of pediatric asthma, including males¹⁶  and non-Hispanic Black children,¹⁷  were more likely to receive a non-bronchiolitis diagnosis. Interestingly, these groups have also been previously described as being at risk for overtreatment with SABAs and corticosteroids in bronchiolitis.^21,22  That previous albuterol use was also associated with more than twice the adjusted odds of a non-bronchiolitis diagnosis may imply that these children have a perceived history of albuterol responsiveness, which some consider to be a distinguishing characteristic between asthma and bronchiolitis.²³  In practice, however, clinicians are caught between recommendations to diagnose bronchiolitis clinically without even a trial of albuterol¹  and recommendations to diagnose asthma on the basis of reversibility of airway obstruction.¹⁸  Some have raised concerns about the recommendation against an albuterol trial for bronchiolitis, particularly in older children or severe disease.^24,25 

These results have important implications for future guidelines for bronchiolitis; assessments of American Academy of Pediatrics bronchiolitis guideline adherence^10,11  and quality improvement efforts to reduce unnecessary resource use^19,26  likely underestimate the true burden of SABA and corticosteroid use by limiting their analysis to children with an ICD-10 code for bronchiolitis. One-quarter of our cohort had a diagnosis other than bronchiolitis, and those patients had greater use of asthma therapies at all ages. Research is needed to help clinicians distinguish the clinical phenotypes of bronchiolitis, asthma, and viral-induced wheezing in the acute setting, particularly in children 12 to 23 months of age. The diagnosis clearly drives the treatment, and heterogeneous and overlapping definitions may limit the applicability of current guidelines.

This study has several limitations. Because PHIS is an administrative database, diagnosis definitions relied on ICD-10 discharge diagnosis codes, which may be subject to variability in billing practices or coding errors. For encounters resulting in inpatient admission from the ED, the PHIS data does not distinguish where discharge diagnoses were made or where treatments occurred, so the evaluation of differences on the basis of location of care is limited. The practice of using a trial of SABA to determine the reversibility of airway obstruction may impact our analysis, particularly because PHIS does not include dosing and frequency of therapeutics. PHIS data are from tertiary children’s hospitals, which limits the generalizability of these results to general EDs, community hospitals, or primary care settings. Although the proportion of pediatric respiratory hospitalizations occurring in children’s hospitals has increased,²⁷  further work should include community settings. The estimates of SABA and steroid use in this study may be underestimates of use across the health system because general EDs have been demonstrated to have increased use of non-recommended treatments for bronchiolitis.²⁸ 

Non-bronchiolitis diagnoses for acute wheezing respiratory illness increase with month of age in children aged 0 to 23 months, as does the use of inhaled bronchodilators and systemic steroids. Hospitals vary widely in their use of diagnosis codes for asthma, bronchiolitis, wheeze, and bronchospasm. Better definitions of clinical phenotypes of bronchiolitis and asthma would allow for more appropriate treatment in acute care settings, particularly in children 12 to 23 months of age.