During this rapidly moving international health crisis, varying and seemingly contrasting reports indicate both a protective and exacerbating contribution of atopy to the severity of coronavirus disease 2019 (COVID‐19) caused by severe acute respiratory syndrome coronavirus‐2 (SARS‐CoV‐2).1, 2 Further, worse outcomes have been reported in association with certain races/ethnicities, highlighting health disparities possibly related to systemic racism, access to care, income inequality, occupational hazards, and disproportionate medical comorbidities.3, 4 Whether these root causes may alter the importance of atopy on COVID‐19 disease course remains unknown. We hypothesized that atopy, as determined by elevated blood eosinophil percent measured at time of hospital admission, would be associated with survival among hospitalized patients, and this protection against mortality may vary across different racial/ethnic groups.
We performed a retrospective analysis of reported demographics, medical history, medications and baseline laboratory values from the electronic medical record (Epic) of 4252 hospitalized patients with a positive SARS‐CoV‐2 polymerase chain reaction (PCR) test in the Mount Sinai Health System from 03/01/20 to 06/07/20 (see Methods section in this article’s Online Repository at https://onlinelibrary.wiley.com/journal/13989995). The following races/ethnicities were included: Asian (n = 202), Black (n = 1141), Hispanic (n = 1152), and White (n = 1004; Table S1). The main outcome of interest was mortality among patients hospitalized for treatment of COVID‐19 across various races/ethnicities, and the secondary outcome was length of hospital stay. We conducted generalized regression models by integrating eosinophil percentages and comorbidities (eg, diabetes, obesity) to predict our two endpoints, survival by a logistic regression model and length of hospital stay by a linear regression model.
We found differential effects of eosinophil percent on patient survival that varied by race/ethnicity using logistic regressions (Table 1). A higher percentage of eosinophils was predictive of higher odds of survival, specifically among Hispanic and White patients (p < .05; Table 1, Figure 1). In other words, fewer eosinophils were associated with higher mortality. Asthma was not a significant predictor of mortality overall, but trended toward lower mortality among Black patients (p < .1; Table 1). Chronic kidney disease (CKD) and obesity were significant predictors of higher mortality among Black patients (p < .01; Table 1). Additionally, older age was predictive of higher mortality across all races/ethnicities (p < .01; Table 1).
Logistic regression model predicting mortality from various variables by race/ethnicity.
| Race | Asian | Black | Hispanic | White | ||||
|---|---|---|---|---|---|---|---|---|
| Variables | Est | Pr (>|z|) | Est | Pr (>|z|) | Est | Pr (>|z|) | Est | Pr (>|z|) |
| Intercept:Mortality | −6.55 | 7.73E−05 | −5.58 | 3.78E−17 | −5.65 | 3.16E−21 | −6.76 | 5.31E−19 |
| Age | 0.07 | 2.06E−04 | 0.05 | 6.73E−13 | 0.05 | 5.17E−18 | 0.07 | 6.33E−19 |
| Asthma | – | – | −0.79 | 7.77E−02 | −0.51 | 1.81E−01 | −0.19 | 6.98E−01 |
| Eosinophil % | −0.58 | 2.32E−01 | −0.13 | 1.73E−01 | −0.31 | 1.64E−02 | −0.34 | 2.89E−02 |
| Sex | 0.37 | 4.58E−01 | 0.19 | 3.08E−01 | −0.06 | 7.44E−01 | 0.31 | 1.10E−01 |
| CKD | −0.09 | 8.95E−01 | 0.83 | 1.07E−03 | 0.28 | 2.82E−01 | 0.15 | 6.64E−01 |
| COPD | −0.12 | 9.24E−01 | 0.72 | 6.74E−02 | −0.29 | 4.39E−01 | 0.08 | 8.26E−01 |
| Diabetes | −0.03 | 9.54E−01 | −0.05 | 8.31E−01 | 0.05 | 8.10E−01 | 0.39 | 1.52E−01 |
| Other immune | – | – | 0.02 | 9.81E−01 | 0.18 | 7.66E−01 | −0.44 | 4.75E−01 |
| Obesity | −14.10 | 9.92E−01 | 0.67 | 2.23E−02 | 0.18 | 5.69E−01 | 0.12 | 7.64E−01 |
| Smoking status | 0.04 | 7.98E−01 | 0.00 | 9.90E−01 | 0.03 | 5.96E−01 | −0.07 | 2.65E−01 |
| Ferritin | 0.00 | 6.12E−01 | 0.00 | 8.04E−01 | 0.00 | 3.52E−02 | 0.00 | 6.74E−04 |
| LDH | 0.00 | 1.74E−01 | 0.00 | 2.11E−04 | 0.00 | 1.14E−08 | 0.00 | 2.74E−04 |
- Abbreviations: CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; LDH, lactate dehydrogenase; Other immune (atopic dermatitis, alopecia areata, ulcerative colitis, Crohn’s disease, rheumatoid arthritis). Values in bold have p<0.05.
Log odds ratio of eosinophil percent of deceased versus survived patients for each race/ethnicity. White and Hispanic patients with a lower eosinophil percent have significant mortality odds. Horizontal bars denote standard error, dots reflect the log odds ratio estimate, *p < .05
The effect of atopy on the length of hospital stay also varied by race/ethnicity. Elevated eosinophil percentage was a significant predictor of shorter hospital stay among Asian patients (p < .05), but was not a significant predictor of hospital stay for White, Black, and Hispanic patients (Table S2). Asthma, age, and other immune conditions (eg, atopic dermatitis, alopecia areata, rheumatoid arthritis, ulcerative colitis, Crohn’s disease) were not found to be significant predictors of hospital stay. Similar to the mortality outcome, CKD was a significant predictor of longer hospital stay among Black patients. COPD was a significant predictor of hospital stay among Asian patients (p < .05; Table S2).
These data reveal the potentially protective role of eosinophils in systemic circulation on mortality and hospital duration. This protection varied across racial/ethnic groups. Limited prior research has evaluated the role of eosinophils on specific disease processes by race/ethnicity. One prior study found elevated levels of eosinophilic airway inflammation among Black asthmatics treated with inhaled corticosteroids (ICS), compared with White asthmatics using ICS.5 The eosinophil percent is particularly important in our study because eosinophils have a role in antiviral immune responses.6 For instance, eosinophils express toll‐like receptors (TLRs) that detect viral microbe‐associated molecular patterns, facilitating cytokine production (eg, IFN‐γ, IL‐12) and nitric oxide generation, which has direct antiviral effects.6
These findings may have implications for the management of COVID‐19 patients. For instance, further investigations are needed to clarify the increasing number of biologic agents that target eosinophils, because decreasing eosinophil counts may potentially contribute to increased mortality among severely ill COVID‐19 patients. Additionally, hospitalized non‐Asian patients with eosinopenia may require additional and more intensive care. From a broader perspective, these data reinforce the necessity to further understand and combat the causes of racial/ethnic inequities in health care outcomes.
Nonetheless, these findings reveal racial and ethnic health disparities in important health outcomes. They also emphasize the potentially protective role of atopy as determined by eosinophils in COVID, and its susceptibility to modification by race/ethnicity. Combined, these findings may provide a greater understanding of how COVID‐19 affects complex immune, social, and environmental components.
ACKNOWLEDGEMENTS
We acknowledge the Mount Sinai Hospital System for obtaining the COVID‐19 registry.
CONFLICTS OF INTEREST
EGY is an employee of Mount Sinai and has received research funds (grants paid to the institution) from: Abbvie, Celgene, Eli Lilly, Janssen, Medimmune/Astra Zeneca, Novartis, Pfizer, Regeneron, Vitae, Glenmark, Galderma, Asana, Innovaderm, Dermira, UCB. EGY is also a consultant for Sanofi Aventis, Regeneron, Stiefel/GlaxoSmithKline, MedImmune, Celgene, Anacor, AnaptysBio, Dermira, Galderma, Glenmark, Novartis, Pfizer, Vitae, Leo Pharma, Abbvie, Eli Lilly, Kyowa, Mitsubishi Tanabe, Asana Biosciences, and Promius. The rest of the authors declare that they have no conflicts of interest to disclose.
