Abstract
Background
Factors responsible for asthma exacerbations in children are complex and may differ from those that drive asthma severity.
Objective
To identify latent classes of children at risk for asthma exacerbation and determine whether latent class assignment is useful in the prediction of future exacerbation.
Methods
Latent class analysis (LCA) was performed on 513 children 6–17 years at risk for asthma exacerbation, with 31 variables encompassing demographics, medical history, treatment, symptoms, lung function, sensitization and Type-2 inflammation. Primary and secondary outcomes included exacerbation occurrence by 12 months and time to first exacerbation, respectively.
Results
Four latent classes were identified with differing demographic features, sensitization and Type-2 inflammatory markers, prior exacerbation severity and healthcare utilization, and lung function. Exacerbations occurred in 22.4% of class 1 (“lesser sensitization with normal lung function”), 27.9% of class 2 (“lesser sensitization with prior severe exacerbation and normal lung function”), 45.3% of class 3 (“multiple sensitization with reversible airflow limitation”), and 64.3% of class 4 (“multiple sensitization with partially reversible airflow limitation”) (p<0.001). Time to exacerbation also followed similar trends and was shortest in the latent classes with multiple sensitization and airflow limitation (p<0.001). Outcomes were driven largely by children with exacerbation prone asthma (defined as ≥3 exacerbations in the prior year), who were present in each class but most strongly represented in classes 3 and 4.
Conclusions
Children at risk for asthma exacerbation are a heterogeneous group. Sensitization, prior exacerbation severity, and lung function variables may be particularly useful in identifying children at greatest risk for future exacerbation.
Keywords: Asthma in children, Phenotype, Asthma exacerbation, Asthma control, Asthma outcomes, Latent class analysis, Lung function, Type-2 inflammation, Aeroallergen sensitization
INTRODUCTION
Asthma control is suboptimal in the majority of children in the United States,1–3 despite widespread availability of asthma controller medications such as inhaled corticosteroids (ICS) and standardized treatment guidelines.4, 5 While deaths from asthma have declined,6 more than half of children with asthma continue to experience an exacerbation each year.7 National surveillance data in children also highlight disturbing trends of decreased primary care visits for asthma and increased emergency department (ED) visits for asthma exacerbations8 that also contribute to school absenteeism,9, 10 decreased productivity in working parents11 and rising asthma costs.11–14
While the factors responsible for asthma exacerbations in children are complex, it is also recognized that children at risk for asthma exacerbations are heterogeneous, with clinical features and pathophysiologic mechanisms that may differ from those that drive asthma severity.15 However, there are very few phenotypic and longitudinal outcome studies of populations of children enriched for high exacerbation risk. Given these knowledge gaps, we applied latent class analysis (LCA) to a dataset of well-characterized children 6–17 years of age with physician diagnosed and confirmed asthma at risk for asthma exacerbation. The purposes of this study were to: (1) identify latent classes of children at risk for asthma exacerbation and (2) determine whether latent class assignment is useful in the prediction of future exacerbation. We hypothesized that a latent class of children with exacerbation-prone asthma (defined as having ≥3 exacerbations in the previous year) would be identified and would be associated with greater symptom burden, lower lung function, poorer quality of life and higher odds of subsequent exacerbation within 12 months of follow-up.
METHODS
Children 6–17 years of age at risk for exacerbation recruited into outpatient asthma clinical research studies at Emory University and Children’s Healthcare of Atlanta between January 1, 2004 and December 31, 2015 were included in the analysis. Criteria for inclusion were: 1) physician-diagnosed asthma with historical evidence of either ≥ 12% reversibility in forced expiratory volume in one second (FEV1) relative to baseline or airway hyperresponsiveness evidenced by a provocative concentration of methacholine causing a 20% drop in FEV1 (PC20) ≤16 mg/mL, and 2) prior asthma exacerbation necessitating treatment with systemic corticosteroids within the preceding 12 months or uncontrolled asthma per treatment guidelines,5 evidenced by daytime symptoms more than two times per week, nocturnal awakenings two or more times per month, or self-reported activity interference. Other exclusions included premature birth before 35 weeks of gestation, current smoking >5 pack years, or other chronic airway disorders such as cystic fibrosis, pulmonary aspiration or vocal cord dysfunction. Permission to proceed with this study was granted by the Emory University and Children’s Healthcare of Atlanta Institutional Review Boards. Informed written consent was obtained from legal guardians. Verbal assent was also obtained from children 6–10 years and written assent was obtained from children and adolescents 11 to 17 years.
Participant recruitment and characterization
Participants were recruited from the ED, hospital wards, and asthma clinics of Children’s Healthcare of Atlanta at Egleston hospital and through community-based advertisements. Participants completed an outpatient study visit during which questionnaires and clinical phenotyping procedures were performed. All participants were stable at the time of the study visit with no self-reported acute illness and no exacerbation treated with systemic corticosteroids within the preceding four weeks. A subset of participants also received a single injection of intramuscular triamcinolone acetonide (1 mg/kg, up to 60 mg maximum dose) for the purpose of systemic corticosteroid responsiveness testing; these patients returned for additional clinical phenotyping procedures at 14 (+7) days.16
Community (i.e. zip code) characteristics were obtained from Tables S101 (Age and Sex), S1501 (Educational Attainment), and DP03 (Selected Economic Characteristics) of the 2010–2014 American Community Survey 5-year estimates.17 Asthma-related quality of life over the preceding two weeks was assessed with the Pediatric Asthma Quality of Life Questionnaire (PAQLQ) that excludes assessment of the environmental domain.18 A subset of participants also completed the Asthma Control Test (ACT),19 the 7-item Asthma Control Questionnaire (ACQ),20 and the 5-item Medication Adherence Report Scale (MARS).21, 22 Exhaled nitric oxide concentrations were measured with a commercial device (NIOX®, Aerocrine, Solna, Sweden or NIOX MINO®, Circrassia Pharmaceuticals, Chicago, IL) according to technical standards.23 Aeroallergen sensitization was assessed by specific IgE testing (Children’s Healthcare of Atlanta, Atlanta, GA) or skin prick testing with 12 extracts: tree mix, grass mix, weed mix, Ambrosia artemisiifolia, Alternaria alternata, Aspergillus fumagatis, Cladosporium herbarum, dog dander, cat dander, Blatella germanica, Dermatophagoides farinae, and Dermatophagoides pteronyssinus (Greer® Laboratories, Lenoir, NC). Venipuncture was also performed for quantification of blood eosinophils (Children’s Healthcare of Atlanta).
Spirometry (KoKo® PDS, Ferraris, Louisville, CO) was performed following a bronchodilator withhold of ≥4 hours for short-acting beta-agonists and ≥12 hours for long-acting beta-agonists. Best FEV1 and forced vital capacity (FVC) values from 3 reproducible maneuvers24 were recorded at baseline and after 360 mcg albuterol sulfate delivered by a metered dose inhaler. Spirometry data were interpreted according to Global Lung Function Initiative prediction equations.25 A subset of participants, on separate days, also underwent lung volume measurement with a body plethysmograph (MedGraphics Elite Series; Medical Graphics Corporation, St Paul, Minn) for quantification of residual volume (RV) and total lung capacity (TLC) and bronchoprovocation testing with methacholine concentrations of 0 to 16 mg/mL (Provocholine; Methapharm Inc, Coral Springs, FL) delivered by a Rosenthal dosimeter (Pulmonary Data Service Instrumentation, Louisville, CO). Bronchoprovocation was limited to participants with baseline FEV1 > 70% predicted. After the study procedures were completed, treatment was adjusted in accordance with treatment guidelines.5
Statistical analyses
LCA was performed using Mplus software v.8 (Muthen & Muthen, Los Angeles, CA) that allows for the use of both continuous and categorical latent variables. The LCA analysis input consisted of 31 variables to determine the optimal number of latent classes, including 10 dichotomous variables, 2 categorical variables, and 19 continuous variables. Dichotomous variables included: (1) sex, (2) Hispanic ethnicity, (3) tobacco smoke exposure (defined as any smoker in any household in which the subject regularly spends time), (4) history of eczema (ever), (5) history of sinusitis (ever), (6) history of pneumonia (ever), (7) indoor pet exposure (defined as a cat or dog inside the home), (8) parent with asthma (ever in lifetime), (9) ED visit for asthma in the past 12 months, and (10) hospitalization for asthma in the past 12 months. Categorical variables included race (black, white, other), and body mass index (BMI) percentile (<85%, 85–95%, ≥95%). Continuous variables included: 1) age, 2) asthma duration, 3) number of asthma controller medications, 4) symptom score over the previous 4 weeks (sum of daytime cough, daytime wheeze, and nighttime symptoms each coded as 0=never, 1=monthly, 2=once weekly, 3=at least twice weekly, 4=daily), 5) baseline FEV1 (% predicted), 6) baseline FVC (% predicted), 7) baseline FEV1/FVC (% predicted), 8) post-bronchodilator FEV1 (% predicted), 9) post-bronchodilator FVC (% predicted), 10) post-bronchodilator FEV1/FVC (% predicted), 11) absolute FEV1 change (defined as post-bronchodilator %predicted – baseline %predicted), 12) exhaled nitric oxide concentration (ppb), 13) serum IgE level (kU/L), 14) blood eosinophil percentage, 15) percentage of positive allergens (out of 12), 16) zip code population, 17) zip code unemployment (%), 18) zip code educational attainment (% with bachelor’s degree), and 19) zip code families below the poverty threshold (%). Models of 1 to 5 latent classes were repeatedly fitted with the number of latent classes in a stepwise fashion. Best fit was assessed via model fit statistics (log-likelihood, Akaike information criterion (AIC), adjusted Bayesian information criterion (BIC)), entropy, the Lo-Mendell-Rubin Adjusted Ratio Test, and clinical judgment. Each participant was assigned to the latent class with the highest membership probability. Features of latent classes were compared with chi-square tests and ANOVA with Tukey’s post-hoc tests.
Outcome analyses
The primary outcome was the proportion of participants within each latent class with any exacerbation necessitating treatment with systemic corticosteroids within 12 months after the study visit,26 assessed by chi-square tests. Logistic regression was used to obtain odds ratios and 95% confidence intervals. The secondary outcome was time to first exacerbation assessed with proportional-hazards regression models. Exploratory outcomes included any exacerbation necessitating hospitalization for asthma and ACQ scores after systemic triamcinolone administration in the subset of participants who underwent corticosteroid responsiveness testing. Analyses utilized a 0.05 significance level without correction for multiple testing given the exploratory nature of the work.
RESULTS
513 children with complete data were included in the LCA. The distribution of study participants by inclusion criterion is shown in Table E1. For the LCA, two, three, four and five class solutions were evaluated. Although the 5-class solution had the best model fits (as indicated by smallest fit-value) and highest entropy (Table E2), this solution yielded an unacceptable sample size (N=3) in one of the resultant latent classes (Table E3). Therefore, the 4-class solution was chosen as best fit; this solution also yielded a high class membership probability for all participants (Figure E1). Individual variables are shown in Figure E2 and represented by z-scores. Demographic, exposure, and medical history characteristics of the four latent classes are shown in Table 1 and symptoms, medications, prior exacerbations, healthcare utilization and Type-2 inflammatory markers are shown in Table 2. Lung function values are shown in Figure 1, air trapping and airway hyperresponsiveness data are shown in Figure 2, and asthma-related quality of life scores are shown in Figure 3. To simplify the discussion, latent classes were assigned a summary label. Key features of the four latent classes are detailed below.
Table 1.
Demographic, exposure, and medical history characteristics of the four latent classes. Data represent the number of participants (%) or the median (25th, 75th percentile).
| Characteristics | Lesser sensitization with normal lung function n=180 | Lesser sensitization with prior severe exacerbation and normal lung function n=143 | Multiple sensitization with reversible airflow limitation n=147 | Multiple sensitization with partially reversible airflow limitation n = 43 |
|---|---|---|---|---|
| Latent class assignment | 1 | 2 | 3 | 4 |
| Age (years) | 11.0 (9.1, 13.4) | 9.42 (7.4, 11.6) | 12.2 (10.0, 14.5) | 13.2 (10.9, 15.6) |
| Asthma duration (years) | 7.7 (5.0, 10.3) | 6.8 (4.0, 8.8) | 10.5 (7.9, 12.9) | 11.0 (8.7, 13.3) |
| Males | 100 (55.6) | 75 (52.4) | 95 (64.6) | 30 (69.8) |
| Race | ||||
| Hispanic | 14 (7.8) | 3 (2.1) | 8 (5.4) | 4 (9.3) |
| Body Mass Index percentile | ||||
| Medical History (ever) | ||||
| Current Exposures | ||||
| Parent with asthma | 85 (47.2) | 72 (50.3) | 96 (65.3) | 30 (69.8) |
| Highest household education1 | ||||
| Zip Code features | ||||
Class 1, n = 136; Class 2, n = 128; Class 3, n = 127; Class 4, n = 35
Table 2.
Symptoms, medications, healthcare utilization and Type-2 inflammatory markers in the four latent classes. Data represent the number of participants (%) or the median (25th, 75th percentile).
| Feature | Lesser sensitization with normal lung function n=180 | Lesser sensitization with prior severe exacerbation and normal lung function n=143 | Multiple sensitization with reversible airflow limitation n= 147 | Multiple sensitization with partially reversible airflow limitation n = 43 |
|---|---|---|---|---|
| Latent class assignment | 1 | 2 | 3 | 4 |
| Symptoms (past 4 weeks) | ||||
| Asthma control | ||||
| Controller Medications | ||||
| Exacerbations (past year) | ||||
| Healthcare utilization | ||||
| Intubation (ever) | 19 (10.6) | 11 (7.7) | 35 (23.8) | 11 (25.6) |
| Type 2 Inflammatory Markers | ||||
| Aeroallergen sensitization | ||||
Class 1, n = 92; Class 2, n = 89; Class 3, n = 78; Class 4, n = 20
Class 1, n = 67; Class 2, n = 74; Class 3, n = 85; Class 4, n = 19
MARS5 score, range 5–25, where higher values reflect greater adherence. Class 1, n = 44; Class 2, n = 41; Class 3, n = 36; Class 4, n = 10
Figure 1.
Pre- and post-bronchodilator FVC, FEV1, and FEV1/FVC in class 1 (lesser sensitization with normal lung function), class 2 (lesser sensitization with prior severe exacerbation and normal lung function), class 3 (multiple sensitization with reversible airflow limitation), and class 4 (multiple sensitization with partially reversible airflow limitation). ap<0.05 vs. class 1, bp<0.05 vs. class 2, cp<0.05 vs. class 3
Figure 2.
(A) RV/TLC and (B) methacholine PC20 concentration in class 1 (lesser sensitization with normal lung function), class 2 (lesser sensitization with prior severe exacerbation and normal lung function), class 3 (multiple sensitization with reversible airflow limitation), and class 4 (multiple sensitization with partially reversible airflow limitation). ap<0.05 vs. class 1, bp<0.05 vs. class 2, cp<0.05 vs. class 3
Figure 3.
Pediatric Asthma Quality of Life Questionnaire (PAQLQ) domain scores in class 1 (lesser sensitization with normal lung function), class 2 (lesser sensitization with prior severe exacerbation and normal lung function), class 3 (multiple sensitization with reversible airflow limitation), and class 4 (multiple sensitization with partially reversible airflow limitation). ap<0.05 vs. class 1, bp<0.05 vs. class 2, cp<0.05 vs. class 3
Lesser sensitization with normal lung function (Latent Class 1)
One hundred eighty children (35.1%) were included in this latent class, termed “lesser sensitization with normal lung function.” This class had the lowest prevalence of co-morbidity and family history of asthma. Children in this class overall had uncontrolled asthma with the fewest controller medications. Twenty-one percent of children in this class (n=38) had exacerbation-prone asthma with ≥3 exacerbations in the prior year, but had lesser healthcare utilization. While the majority of these children (~75%) had sensitization to at least one aeroallergen Type-2 inflammatory features tended to be lower in this group. Lung function was largely preserved in this latent class. This latent class also had minimal air trapping, lesser airway hyperresponsiveness to methacholine and the greatest quality of life.
Lesser sensitization with prior severe exacerbation and normal lung function (Latent Class 2)
One hundred forty three children (27.9%) were included in this latent class, termed “lesser sensitization with prior severe exacerbation and normal function.” Children in this latent class were younger and tended to be more obese. Symptoms, controller medications, prior exacerbations, sensitization patterns and Type-2 inflammatory features were similar to those of children with lesser sensitization and normal lung function. Twenty one percent (n=30) of children in this class had ≥3 exacerbations in the prior year, but unlike children with lesser sensitization and normal lung function, these children had more unscheduled visits and hospitalizations for asthma. Baseline FVC and FEV1 values were also higher in this group, but FEV1/FVC ratios and quality of life measures were similar to those of children children with lesser sensitization and normal lung function.
Multiple sensitization with reversible airflow limitation (Latent Class 3)
One hundred forty seven children (28.7%) were included in this latent class, termed “multiple sensitization and reversible airflow limitation.” Children in this latent class were predominantly black males with a higher prevalence of parental asthma and comorbidity. These children also resided in more disadvantaged neighborhoods and had uncontrolled asthma with frequent symptoms, more prior exacerbations, indoor tobacco smoke exposure, multiple sensitization, and the highest magnitude of Type-2 inflammatory markers despite more controller medications. However, medication adherence scores were also lowest suggesting lesser adherence to ICS. Thirty four percent of children in this latent class (n=50) met criteria for exacerbation-prone asthma, with ≥3 exacerbations in the prior year. This latent class was also distinguished by a high prevalence of near-fatal asthma necessitating intubation, frequent healthcare utilization in the prior year, and mild yet reversible airflow limitation accompanied by mild air trapping, greater airway hyperresponsiveness to methacholine, and more impaired quality of life.
Multiple sensitization with partially reversible airflow limitation (Latent Class 4)
Forty three children (8.3%) were included in this latent class, termed “multiple sensitization with partially reversible airflow limitation.” Children in this latent class were older, mostly black males with a higher prevalence of parental asthma and comorbidity such as pneumonia. Children in this latent class also resided in more disadvantaged neighborhoods and had the poorest asthma control, highest exacerbation rate before study enrollment, more prevalent indoor tobacco smoke exposure, and most frequent healthcare utilization in the prior year. Forty seven percent of children in this latent class (n=20) met criteria for exacerbation-prone asthma, with ≥3 exacerbations in the prior year. Children in this latent class also had a high prevalence of lifetime intubation for asthma, multiple sensitization, and elevated Type-2 inflammatory markers despite more controller medications. Baseline lung function was lowest in this group and was accompanied by significant air trapping and the greatest airway hyperresponsiveness to methacholine. Airflow limitation was only partially reversible with bronchodilation and was also accompanied by more impaired quality of life.
Exacerbation Outcomes
The primary outcome, any exacerbation occurrence within 12 months after the study visit, was assessed from 494 participants (96.3%). Exacerbations occurred in 22.4% of class 1 (“lesser sensitization with normal lung function”), 27.9% of class 2 (“lesser sensitization with prior severe exacerbation and normal lung function”), 45.3% of class 3 (“multiple sensitization with reversible airflow limitation”), and 64.3% of class 4 (“multiple sensitization with partially reversible airflow limitation”) (p<0.001) (Figure 4A, Table 3). Outcomes were almost completely driven by children with exacerbation prone asthma, defined as ≥3 exacerbations in the previous year (Table 3).
Figure 4.
(A) Exacerbation (solid bar) and exacerbation with hospitalization (dashed bar) occurrence by 12 months, (B) time to first exacerbation, and (C) Asthma Control Questionnaire (ACQ) scores after triamcinolone administration, by latent class assignment. ap<0.05 vs. class 1, bp<0.05 vs. class 2, cp<0.05 vs. class 3
Table 3.
Odds of exacerbation by 12 months according to latent class assignment.
| Exacerbation outcome | N with outcome data | N (%) with exacerbation | Odds ratio (95% confidence interval) | |
|---|---|---|---|---|
| All participants | ||||
| Class 1 | Lesser sensitization with normal lung function | 174 | 39 (22.4) | -- |
| Class 2 | Lesser sensitization with prior severe exacerbation and normal lung function | 136 | 38 (27.9) | 1.34(0.80,2.25) |
| Class 3 | Multiple sensitization with reversible airflow limitation | 139 | 63 (45.3) | 2.87(1.76,4.68) |
| Class 4 | Multiple sensitization with partially reversible airflow limitation | 42 | 27 (64.3) | 6.23(3.01, 12.82) |
| Participants with exacerbation-prone asthma2 | ||||
| Class 1 | Lesser sensitization with normal lung function | 36 | 13(36.1) | -- |
| Class 2 | Lesser sensitization with prior severe exacerbation and normal lung function | 45 | 18(64.3) | 3.19(1.14,8.92) |
| Class 3 | Multiple sensitization with reversible airflow limitation | 28 | 32(71.1) | 4.36(1.71, 11.12) |
| Class 4 | Multiple sensitization with partially reversible airflow limitation | 20 | 17(85.0) | 10.03(2.46,40.79) |
Defined as requiring treatment with systemic corticosteroids
Defined as ≥3 exacerbations in prior year
Time to exacerbation, adjusted for season of enrollment, was also shorter for children with multiple sensitization and reversible airflow limitation (latent class 3, adjusted hazard ratio 2.51, 95% CI: 1.68, 3.75) and children with multiple sensitization and partially reversible airflow limitation (latent class 4, adjusted hazard ratio 3.87, 95% CI: 2.36, 6.36) (Figure 4B). In exploratory analyses, children in Classes 3 and 4 also had a higher occurrence of hospitalization (15% and 29%, respectively) (Figure 4A). However, in the subset of children who underwent systemic corticosteroid responsiveness testing with intramuscular triamcinolone, the majority of children in Class 3 (n=17, 65%) achieved “well controlled” asthma with triamcinolone administration, evidenced by an ACQ score <0.75.27 In contrast, a significant proportion of children in Class 4 (n=8, 57%) had “poorly controlled” asthma after triamcinolone, evidenced by an ACQ score >1.2528 (Figure 4C).
DISCUSSION
In this study, we used LCA to uncover previously unobservable patterns in a well-characterized study population of children at risk for exacerbation. LCA was selected over other exploratory approaches because it is model-based, allows model comparisons to be statistically tested, and is appropriate for questionnaire-derived data yet permits differing variable measurement scales and variances.29 Using LCA, we identified two latent classes of children with a lesser magnitude of sensitization and normal lung function who differed with regard to prior exacerbation severity. Approximately 21% of children in each of these classes met criteria for exacerbation prone asthma with ≥3 exacerbations in the previous year, and nearly 25% of children in each of these two classes had a subsequent exacerbation within 12 months. In contrast, we also identified two latent classes of children with multiple sensitization who differed with regard to baseline lung function and the degree of bronchodilator reversibility. These classes had a higher proportion of participants with exacerbation prone asthma and subsequent exacerbations were nearly 2- to 4-fold higher in these classes. These exacerbations also tended to be more severe and were accompanied by greater hospitalization for status asthmaticus.
Consistent with our hypothesis, we did identify two latent classes with more symptom burden and lower lung function parameters and impaired quality of life at highest risk for subsequent exacerbation. However, exacerbations did occur in all four latent classes. Furthermore, participants with exacerbation-prone asthma (defined as having ≥3 exacerbations in the previous year) were also present in each latent class. Therefore, we cannot say with certainty that these latent classes conform to distinct “phenotypes” of exacerbation-prone asthma. Indeed, the LCA approach is considered exploratory and hypothesis-generating. Nonetheless, our findings do have clinical and biological plausibility. Other cluster analyses of children with more severe asthma (who also tend to have frequent exacerbations) have likewise identified multiple clusters differentiated by the degree of allergic sensitization, asthma duration, controller medications, body mass index and pre-bronchodilator lung function.30–35 A consistent theme across these analyses is greater severity of asthma in clusters with more prominent Type-2 inflammatory features, but some Type-2 features are present in nearly all the pediatric clusters that have been described. For example, in children 6–11 years with difficult-to-treat asthma, Schatz et al. identified a moderately-prevalent (31%) cluster of children without allergic rhinitis or atopic dermatitis who also had significant airflow obstruction.34 However, mean IgE levels were also elevated in this group, suggesting that at least some children had underlying Type-2 inflammation.34 Similarly, a cluster analysis of inner-city children with asthma also identified a low-prevalent (15%) cluster with the lowest levels blood eosinophils and exhaled nitric oxide, but several children in this group had aeroallergen mono-sensitization and 28% had a positive Phadiatop allergen sensitization screen.35
Our observation of increased exacerbation occurrence in the two latent classes of children with multiple sensitization and airflow limitation is also similar to other findings from hypothesis-directed research. Multiple regression analyses have identified several independent risk factors of subsequent asthma exacerbations in children, including race and socioeconomic variables,36 the magnitude of airflow limitation and reversal with bronchodilation,37 multiple-versus mono-sensitization to aeroallergens,38 and a prior exacerbation requiring treatment with systemic corticosteroids.39 Body mass index has also been implicated as a risk factor for exacerbation-prone asthma in children (defined as ≥3 exacerbation in the previous year)15 and, in meta-analyses, was associated with a small yet significant increased risk of exacerbation in children independent of sex.40 However, very few studies have examined associations between multiple variable groupings (i.e., clusters or latent classes) and exacerbation outcomes. For example, the cluster analysis of inner-city children with asthma identified a cluster with the highest degree of Type-2 inflammation and multiple aeroallergen sensitization (~15% of children) that also had the highest medication requirements, greatest airflow obstruction and airway lability, and the highest proportion of exacerbations requiring treatment with systemic corticosteroids.35 However, the prospective rate of exacerbations was included as a variable in that clustering algorithm and not assessed independently of cluster assignment.35 A cluster analysis of the Childhood Asthma Management Research Program (which excludes severe asthma) did identify a highly atopic cluster of children with airflow limitation, the highest bronchodilator response, severe airway hyperresponsiveness and the highest baseline exacerbation rate who also had the shortest time to systemic corticosteroids for asthma exacerbation over four years of follow up.31 Furthermore, 64% of participants in this group received systemic corticosteroids within the first year.26 However, that study was limited to children with mild-to-moderate asthma and excluded children with more severe forms of asthma and frequent exacerbations.
Strengths of the present study include the relatively large and racially diverse sample of children at risk for exacerbation who underwent extensive phenotypic characterization prior to longitudinal study. However, there are a number of limitations. Admittedly, the factors responsible for asthma exacerbations in children are complex and include limited access to primary care41 and exposures to environmental allergens and irritants in addition to sensitization42–44 that were not adequately addressed by the study design. Poor adherence to controller medications is also common in general populations and was not thoroughly addressed by the study design.3, 45 Although MARS scores for medication adherence suggest that adherence was acceptable in Classes 1, 2 and 4, this score may overestimate adherence when assessed by electronic means.46 Although asthma controller therapies were adjusted at the completion of the baseline characterization visit, it is therefore possible that poor adherence to ICS impacted the exacerbation outcomes. Indeed, the exploratory analysis in a limited subset of participants undergoing systemic corticosteroid responsiveness testing with intramuscular triamcinolone suggests that exacerbations may be largely preventable in Class 3, but less so in Class 4. Similarly, in the cluster analysis performed by the National Heart Lung and Blood Institute’s Severe Asthma Research Program, children with the earliest age of symptom onset and more advanced, partially reversible airflow limitation also had the greatest burden of symptoms and associated medication use at the baseline visit,30 but also had best treatment responses to fluticasone/salmeterol in highly supervised, separate replication analyses.47 However, a second cluster that also had early onset disease and airflow limitation had the least response to guideline-based asthma treatment in that same study.47 Similarly, the cluster of children with low lung function and the highest baseline exacerbation rate identified by the Childhood Asthma Management Research Program also did not improve with daily budesonide treatment.31 Other limitations include the single- versus multi-center design and recruitment of children from an academic medical center, which prohibits generalization to other populations. The number of participants who underwent baseline phenotyping also limited the clustering solution to four groups, which may not be sufficient for detection of low-prevalent asthma phenotypes. The study design also prohibited assessment of the temporal stability of latent classes and transition over time.
In conclusion, using LCA, we identified four latent classes of children at risk for asthma exacerbation that differed with regard to demographic and medical history variables, symptoms, medications, healthcare utilization, Type-2 inflammatory markers and lung function. Children with exacerbation-prone asthma were present in each latent class, but were most strongly represented in the latent classes with multiple sensitization and airflow limitation. These same latent classes had the greatest disease burden and poorest quality of life at baseline and the highest odds of exacerbation by 12 months. Additional studies are needed to determine whether these latent classes correspond to clinically useful phenotypes.
Supplementary Material
HIGHLIGHTS BOX.
What is already known about this topic? (word count = 35)
Children at risk for asthma exacerbations are a heterogeneous group, with clinical features and pathophysiologic mechanisms that may differ from those that drive asthma severity. There are few studies of children at risk for exacerbation.
What does this article add to our knowledge? (word count = 33)
Latent class analysis identified four groups that differed in demographic features, sensitization/Type-2 markers, lung function variables, and future exacerbation occurrence. Exacerbations were most prevalent in latent classes with multiple sensitization and airflow limitation.
How does this study impact current management guidelines? (word count = 35)
Multiple sensitization, in addition to previous exacerbations and lung function, may be a useful predictor of future exacerbations. However, exacerbations were common in each latent class and highlight the heterogeneity of asthma in school-age children.
ACKNOWLEDGMENTS
We acknowledge the Emory+Children’s Pediatric Research Biostatistics Core for assistance with statistical analyses.
Grant support:
This work was supported by R01NR013700, UL1TR002378, K12HD072245, K24AT009893, and the Children’s Healthcare of Atlanta Pediatric Research Alliance Center for Clinical Outcomes and Public Health Research
ABBREVIATIONS
- ACT
Asthma Control Test
- ACQ
Asthma Control Questionnaire
- AIC
Akaike information criterion
- BIC
Bayesian information criterion
- ED
Emergency department
- FEV1
Forced expiratory volume in one second
- FVC
Forced vital capacity
- ICS
Inhaled corticosteroid
- LCA
Latent class analysis
- MARS
Medication Adherence Report Scale
- PAQLQ
Pediatric Asthma Quality of Life Questionnaire
- PC20
Provocative concentration of methacholine causing a 20% decline in FEV1
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Disclosure statement:
Dr. Grunwell, Mr. Gillespie, and Dr. Fitzpatrick have nothing to disclose. Dr. Morris is the inventor or co-inventor of several UCSF-Benioff Children’s Hospital Oakland patents/patent-pending applications that include nutritional supplements, and biomarkers of cardiovascular disease, is an inventor of an Emory University School of Medicine patent application for a nutritional supplement, is a consultant for Pfizer, and has received research support from MAST Therapeutics unrelated to the present work.
REFERENCES
- 1.Szefler SJ. Advancing asthma care: the glass is only half full! J Allergy Clin Immunol 2011; 128:485–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Mensah GA, Kiley JP, Gibbons GH. Generating evidence to inform an update of asthma clinical practice guidelines: Perspectives from the National Heart, Lung, and Blood Institute. J Allergy Clin Immunol 2018. [DOI] [PubMed] [Google Scholar]
- 3.Slejko JF, Ghushchyan VH, Sucher B, Globe DR, Lin SL, Globe G, et al. Asthma control in the United States, 2008–2010: indicators of poor asthma control. J Allergy Clin Immunol 2014; 133:1579–87. [DOI] [PubMed] [Google Scholar]
- 4.Global Initiative for Asthma. Global Strategy for Asthma Management and Prevention, 2018. Available from: www.ginasthma.org. Last accessed: August 13, 2018.
- 5.National Asthma E, Prevention P. Expert Panel Report 3 (EPR-3): Guidelines for the Diagnosis and Management of Asthma-Summary Report 2007. J Allergy Clin Immunol 2007; 120:S94–138. [DOI] [PubMed] [Google Scholar]
- 6.Moorman JE, Akinbami LJ, Bailey CM, Zahran HS, King ME, Johnson CA, et al. National surveillance of asthma: United States, 2001–2010. Vital Health Stat 3 2012:1–58. [PubMed] [Google Scholar]
- 7.Centers for Disease Control and Prevention. Most recent national asthma data: National prevalence of asthma attacks (2017). Available at http://cdc.gov/asthma/most_recent_national_asthma_data.htm. Last accessed February 4, 2020.
- 8.Kaur BP, Lahewala S, Arora S, Agnihotri K, Panaich SS, Secord E, et al. Asthma: Hospitalization Trends and Predictors of In-Hospital Mortality and Hospitalization Costs in the USA (2001–2010). Int Arch Allergy Immunol 2015; 168:71–8. [DOI] [PubMed] [Google Scholar]
- 9.Linneberg A, Dam Petersen K, Hahn-Pedersen J, Hammerby E, Serup-Hansen N, Boxall N. Burden of allergic respiratory disease: a systematic review. Clin Mol Allergy 2016; 14:12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Nurmagambetov T, Khavjou O, Murphy L, Orenstein D. State-level medical and absenteeism cost of asthma in the United States. J Asthma 2017; 54:357–70. [DOI] [PubMed] [Google Scholar]
- 11.Sullivan PW, Slejko JF, Ghushchyan VH, Sucher B, Globe DR, Lin SL, et al. The relationship between asthma, asthma control and economic outcomes in the United States. J Asthma 2014; 51:769–78. [DOI] [PubMed] [Google Scholar]
- 12.Szefler SJ, Zeiger RS, Haselkorn T, Mink DR, Kamath TV, Fish JE, et al. Economic burden of impairment in children with severe or difficult-to-treat asthma. Ann Allergy Asthma Immunol 2011; 107:110–9 e1. [DOI] [PubMed] [Google Scholar]
- 13.Barnett SB, Nurmagambetov TA. Costs of asthma in the United States: 2002–2007. J Allergy Clin Immunol 2011; 127:145–52. [DOI] [PubMed] [Google Scholar]
- 14.Nurmagambetov T, Kuwahara R, Garbe P. The Economic Burden of Asthma in the United States, 2008–2013. Ann Am Thorac Soc 2018; 15:348–56. [DOI] [PubMed] [Google Scholar]
- 15.Denlinger LC, Phillips BR, Ramratnam S, Ross K, Bhakta NR, Cardet JC, et al. Inflammatory and Comorbid Features of Patients with Severe Asthma and Frequent Exacerbations. Am J Respir Crit Care Med 2017; 195:302–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Fitzpatrick AM, Stephenson ST, Brown MR, Nguyen K, Douglas S, Brown LAS. Systemic Corticosteroid Responses in Children with Severe Asthma: Phenotypic and Endotypic Features. J Allergy Clin Immunol Pract 2017; 5:410–9 e4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.United States Census Bureau. 2010–2014 American Community Survey 5-Year Estimates. Available at http://www.census.gov/acs/www/data/data-tables-and-tools/.Last accessed: June 3, 2019.
- 18.Juniper EF, Guyatt GH, Feeny DH, Ferrie PJ, Griffith LE, Townsend M. Measuring quality of life in children with asthma. Qual Life Res 1996; 5:35–46. [DOI] [PubMed] [Google Scholar]
- 19.Nathan RA, Sorkness CA, Kosinski M, Schatz M, Li JT, Marcus P, et al. Development of the asthma control test: a survey for assessing asthma control. J Allergy Clin Immunol 2004; 113:59–65. [DOI] [PubMed] [Google Scholar]
- 20.Juniper EF, O’Byrne PM, Guyatt GH, Ferrie PJ, King DR. Development and validation of a questionnaire to measure asthma control. Eur Respir J 1999; 14:902–7. [DOI] [PubMed] [Google Scholar]
- 21.Cohen JL, Mann DM, Wisnivesky JP, Home R, Leventhal H, Musumeci-Szabo TJ, et al. Assessing the validity of self-reported medication adherence among inner-city asthmatic adults: the Medication Adherence Report Scale for Asthma. Ann Allergy Asthma Immunol 2009; 103:325–31. [DOI] [PubMed] [Google Scholar]
- 22.Chan AHY, Horne R, Hankins M, Chisari C. The Medication Adherence Report Scale (MARS-5): a measurement tool for eliciting patients’ reports of non-adherence. Br J Clin Pharmacol 2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.American Thoracic S, European Respiratory S. ATS/ERS recommendations for standardized procedures for the online and offline measurement of exhaled lower respiratory nitric oxide and nasal nitric oxide, 2005. Am J Respir Crit Care Med 2005; 171:912–30. [DOI] [PubMed] [Google Scholar]
- 24.Miller MR, Hankinson J, Brusasco V, Burgos F, Casaburi R, Coates A, et al. Standardisation of spirometry. Eur Respir J 2005; 26:319–38. [DOI] [PubMed] [Google Scholar]
- 25.Quanjer PH, Stanojevic S, Cole TJ, Baur X, Hall GL, Culver BH, et al. Multi-ethnic reference values for spirometry for the 3–95-yr age range: the global lung function 2012 equations. Eur Respir J 2012; 40:1324–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Fuhlbrigge A, Peden D, Apter AJ, Boushey HA, Camargo CA Jr., Gern J, et al. Asthma outcomes: exacerbations. J Allergy Clin Immunol 2012; 129:S34–48. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Juniper EF, Bousquet J, Abetz L, Bateman ED, Committee G. Identifying ‘well-controlled’ and ‘not well-controlled’ asthma using the Asthma Control Questionnaire. Respir Med 2006; 100:616–21. [DOI] [PubMed] [Google Scholar]
- 28.Nguyen JM, Holbrook JT, Wei CY, Gerald LB, Teague WG, Wise RA, et al. Validation and psychometric properties of the Asthma Control Questionnaire among children. J Allergy Clin Immunol 2014; 133:91–7 e1–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Collins LM, Lanza ST. Latent class and latent transition analysis: With applications in the social, behavioral, and health sciences. Hoboken, NJ: John Wiley & Sons, Inc; 2010. [Google Scholar]
- 30.Fitzpatrick AM, Teague WG, Meyers DA, Peters SP, Li X, Li H, et al. Heterogeneity of severe asthma in childhood: confirmation by cluster analysis of children in the National Institutes of Health/National Heart, Lung, and Blood Institute Severe Asthma Research Program. J Allergy Clin Immunol 2011; 127:382–9 e1–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Howrylak JA, Fuhlbrigge AL, Strunk RC, Zeiger RS, Weiss ST, Raby BA, et al. Classification of childhood asthma phenotypes and long-term clinical responses to inhaled anti-inflammatory medications. J Allergy Clin Immunol 2014; 133:1289–300, 300 e1–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Just J, Gouvis-Echraghi R, Rouve S, Wanin S, Moreau D, Annesi-Maesano I. Two novel, severe asthma phenotypes identified during childhood using a clustering approach. Eur Respir J 2012; 40:55–60. [DOI] [PubMed] [Google Scholar]
- 33.Just J, Saint-Pierre P, Gouvis-Echraghi R, Laoudi Y, Roufai L, Momas I, et al. Childhood allergic asthma is not a single phenotype. J Pediatr 2014; 164:815–20. [DOI] [PubMed] [Google Scholar]
- 34.Schatz M, Hsu JW, Zeiger RS, Chen W, Dorenbaum A, Chipps BE, et al. Phenotypes determined by cluster analysis in severe or difficult-to-treat asthma. J Allergy Clin Immunol 2014; 133:1549–56. [DOI] [PubMed] [Google Scholar]
- 35.Zoratti EM, Krouse RZ, Babineau DC, Pongracic JA, O’Connor GT, Wood RA, et al. Asthma phenotypes in inner-city children. J Allergy Clin Immunol 2016; 138:1016–29. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Fitzpatrick AM, Gillespie SE, Mauger DT, Phillips BR, Bleecker ER, Israel E, et al. Racial disparities in asthma-related health care use in the National Heart, Lung, and Blood Institute’s Severe Asthma Research Program. J Allergy Clin Immunol 2019; 143:2052–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Pongracic JA, Krouse RZ, Babineau DC, Zoratti EM, Cohen RT, Wood RA, et al. Distinguishing characteristics of difficult-to-control asthma in inner-city children and adolescents. J Allergy Clin Immunol 2016; 138:1030–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Beck AF, Huang B, Kercsmar CM, Guilbert TW, McLinden DJ, Lierl MB, et al. Allergen sensitization profiles in a population-based cohort of children hospitalized for asthma. Ann Am Thorac Soc 2015; 12:376–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Chipps BE, Zeiger RS, Borish L, Wenzel SE, Yegin A, Hayden ML, et al. Key findings and clinical implications from The Epidemiology and Natural History of Asthma: Outcomes and Treatment Regimens (TENOR) study. J Allergy Clin Immunol 2012; 130:332–42 e10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Ahmadizar F, Vijverberg SJ, Arets HG, de Boer A, Lang JE, Kattan M, et al. Childhood obesity in relation to poor asthma control and exacerbation: a meta-analysis. Eur Respir J 2016; 48:1063–73. [DOI] [PubMed] [Google Scholar]
- 41.Kodjebacheva GD, Sabo T, Parker S. Influences of asthma on reported health indicators and access to health care among children. Ann Allergy Asthma Immunol 2016; 116:126–33. [DOI] [PubMed] [Google Scholar]
- 42.Wang Z, May SM, Charoenlap S, Pyle R, Ott NL, Mohammed K, et al. Effects of secondhand smoke exposure on asthma morbidity and health care utilization in children: a systematic review and meta-analysis. Ann Allergy Asthma Immunol 2015; 115:396–401 e2. [DOI] [PubMed] [Google Scholar]
- 43.Sala KA, Carroll CL, Tang YS, Aglio T, Dressler AM, Schramm CM. Factors associated with the development of severe asthma exacerbations in children. J Asthma 2011; 48:558–64. [DOI] [PubMed] [Google Scholar]
- 44.Dick S, Doust E, Cowie H, Ayres JG, Turner S. Associations between environmental exposures and asthma control and exacerbations in young children: a systematic review. BMJ Open 2014; 4:e003827. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Kit BK, Simon AE, Ogden CL, Akinbami LJ. Trends in preventive asthma medication use among children and adolescents, 1988–2008. Pediatrics 2012; 129:62–9. [DOI] [PubMed] [Google Scholar]
- 46.Garcia-Marcos PW, Brand PL, Kaptein AA, Klok T. Is the MARS questionnaire a reliable measure of medication adherence in childhood asthma? J Asthma 2016; 53:1085–9. [DOI] [PubMed] [Google Scholar]
- 47.Chang TS, Lemanske RF Jr., Mauger DT, Fitzpatrick AM, Sorkness CA, Szefler SJ, et al. Childhood asthma clusters and response to therapy in clinical trials. J Allergy Clin Immunol 2014; 133:363–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.