Asthma phenotypes and COVID-19 risk
In a univariate analysis, asthma was significantly associated with hospitalization for COVID-19 (Figure 2A and 2B, unadjusted; OR 1.53 [95% CI 1.2, 1.93], p<0.001). This association remained significant after adjusting for age, gender, and ethnicity (Figure 2B, adjusted for demographics; OR 1.55 [95% CI 1.21, 1.97], p<0.001). However, after we adjusted for known comorbidities associated with more severe disease (i.e., diabetes, obesity, coronary heart disease, and hypertension),18,19 asthma was no longer a significant risk factor for hospitalization (Figure 2B, adjusted for demographics and coexisting conditions; OR 1.12 (95% CI 0.86, 1.45), p=0.40). Twenty percent of hospitalized asthmatics were critical, compared to 21% of non-asthmatic patients (Figure 3A). Additionally, asthma was not significantly associated with COVID-19 disease severity among hospitalized inpatients in univariate (Figure 3B, unadjusted model; OR 1.14 (0.78, 1.67), p=0.48) or multivariate analysis when adjusted for age, race, sex and common comorbidities (Figure 3B, adjusted for demographics; OR 1.28 (0.84, 1.95), p=0.24; adjusted for demographics and coexisting conditions; OR 1.21 (0.8, 1.85), p=0.37).
In our asthmatic cohort, 167 patients (27.9%: 148 outpatients, 19 inpatients) were identified as having an allergic asthma phenotype (Figure 4A). Allergic asthma was significantly associated with a lower risk of hospitalization as compared to non-allergic asthma in both univariate (Figure 4B, unadjusted; OR 0.55 (0.31, 0.92), p=0.029) and multivariate analysis (Figure 4B, adjusted for demographics; OR 0.54 (0.3, 0.93), p=0.031); adjusted for demographics and coexisting conditions; OR 0.52 (0.28, 0.91), p=0.026). For hospitalized patients, allergic asthma status did not correlate with disease severity in any of the analyses (Figures 4C and 4D).
Eosinophil counts prior to COVID-19 infection (historical eosinophils), an allergic biomarker, were available for 2,070 patients (398 inpatients and 1,672 outpatients) and significantly higher (Kruskal-Wallis test, p<0.0001) in asthmatics (Supplemental Figure 2). Furthermore, when assessing absolute eosinophil counts (AECs) during hospitalization in patients without steroid, those with severe COVID-19 disease had lower eosinophil levels during hospitalization as compared to patients with less severe disease, independent of asthma status (p=0.0014, Figure 5).
To further understand asthma phenotypes, we assessed outpatient controller or rescue medication use for SARS-CoV-2 positive asthmatics with available records (368 patients, 61.6%: 307 outpatients, 61 inpatients); 213 patients were prescribed GINA step 1-2 therapies (57.9%). Among COVID-19 hospitalized asthmatic inpatients, the majority of patients were prescribed GINA step 1 therapies (n=25, 41%); four patients (6.6%) were prescribed GINA step 5 therapies. In asthmatics, there was no association between GINA classification and the risk for hospitalization (p=0.22) or severity of illness (p=0.62) (Supplemental Figure 3A and 3B).
Non-allergic phenotypes of asthma were also assessed.  There were 32 patients (14 hospitalized and 18 not hospitalized) in our cohort with a diagnosis of asthma/COPD overlap. Patients with a diagnosis of both asthma/COPD were more likely to be hospitalized in univariate analysis (Supplemental Figure 4B - unadjusted; OR 5.71 (2.77, 11.52), p<0.0001) and multivariate analysis adjusted for demographics (OR 3.08 (1.46, 6.4), p=0.0026). However, this relationship did not persist for either group in multivariate analysis after adjusting for demographics and comorbid conditions (OR 1.69 (0.76, 3.7), p=0.19). For hospitalized patients, those with COPD only showed significantly increased odds of clinical severity in univariate analysis (Supplemental Figure 4D - unadjusted; OR 3.9 (1.75, 9.02), p=0.001), which did not persist when adjusting for demographics and comorbid conditions in multivariate analysis (p=0.15).