Results
Age and gender distribution for all included hospitalized COVID-19 cohorts stratified for the presence of asthma are shown in Table 1. The participating German centers included asthmatics who were significantly younger than the non-asthmatic group, while Stanford asthma group had significantly more male patients (Table 1, p-value = 0.0343). The vast majority (> 90%) of patients included in the German and Russian cohorts were Caucasian, albeit we could not collect precise data on ethnicity for these cohorts. The US cohort comprised of 246 (50.7%) Hispanics and 18 (3.7%) Afro-Americans. Asthma was significantly underrepresented (vs. prevalence among adults in local community) in hospitalized COVID-19 cohorts of all included countries in our study with the exception of the Stanford cohort. The latter encompassed asthmatics with a prevalence of 18.35% as compared to a 10.56% prevalence of asthma in the broader California area (Figure 1). We assessed the prevalence of asthma among intensive care unit (ICU) patients and found that it did not significantly differ from the prevalence among patients in normal care for any of the participating centers (Supplementary Figure 1).
We next examined the presence of comorbidities among hospitalized COVID-19 patients across our centers and found that the Stanford cohort exhibits an over-representation of asthma & COPD patients versus non-asthma & COPD patients (Figure 2a, p-value = 0.0046). We observed a similar trend for other comorbidities (concurrent or past) such as cancer and chronic renal disease for patients hospitalized in Stanford, however these differences between asthma and non-asthma hospitalized COVID-19 patients did not reach statistical significance (Figure 2a). Furthermore, the patients in Stanford had more (total) comorbidities compared to Germany and Moscow (Figure 2a). Importantly, the asthma group in Stanford had more additional preconditions than the non-asthma group with over 85% of asthmatics having an additional comorbidity (Figure 2b, p-value = 0.0346). This was not the case with the German and Moscow centers, where asthmatics and non-asthmatics showed a similar pattern in terms of frequency of additional comorbidities (Figure 2b, Germany: p=0.216, Moscow: p=0.8256). Furthermore, a second ‘wave’ of comorbidity frequency was recorded with a second peak after 3 comorbidities on top of asthma and COVID-19 (Figure 2b). The overrepresentation of asthmatics among hospitalized COVID-19 patients in Stanford can be explained by confounders, like age, sex, and comorbidities. To mitigate the effect of these confounders we performed logistic regression to predict asthma using the center (“moscow” or “stanford”, where microdata was available), sex and the 11 comorbities. This resulted in a decrease in the Odds to be asthmatic given that the center was “stanford”, the sex was “female” and no comorbidity was present in all four age groups (Figure 3). The 95% confidence intervals of so-adjusted Odds reach the population level and lower Odds than the population level are not excluded.
We next analyzed basic lab values of all included patients across study centers and observed a peripheral blood eosinopenia at admission in all centers except Stanford, followed by a recovery close to discharge (Figure 4, Supplementary Figure 2). The Stanford group showed higher levels at admission and overlapping values until discharge i.e., no significant change throughout their hospitalization. Platelet counts showed a somewhat similar pattern; however, both asthma and non-asthma patients at Stanford had relatively stable counts throughout. Values of all other studied laboratory parameters did not significantly deviate between centers of our network (Supplementary Figure 2).