RESUMO
Background: Corticosteroid dosing in COVID-19 cases associated with early-onset and late-onset hypoxia have not been separately explored. Methods: In this retrospective cohort study, we divided hypoxic COVID-19 cases into groups based on timing of initiation of corticosteroids relative to onset of symptoms; Group A (≤6th day), Group B (7th-9th day) and Group C (≥10th day), each group being sub-grouped into high and low-to-moderate dose corticosteroid recipients. Cox regression with propensity scoring was used to compare 28-day mortality between high and low-to-moderate dose recipients separately in Group A, Group B, Group C. Results: Among 505 patients included, propensity score matched Cox regression showed greater risk of all-cause mortality among high dose recipients in Group A [HR= 7.35, 95%CI 3.36-16.11, p-value<0·01, N=114] and Group B [HR=3.17, 95%CI 1.65-6.07, p-value<0·01, N=251]. In Group C, mortality was lowest [12.8% (18/140)] with no significant difference between sub-groups [HR=2.52, 95%CI 0.22-29.15, p-value=0.459, N=140]. Kruskal-Wallis Test between Group A, Group B and Group C for six pre-defined exposure variables showed significant differences for Neutrophil:Lymphocyte Ratio (NLR). Conclusion: When steroids were initiated early (owing to an earlier onset of hypoxic symptoms), a high dose of corticosteroid was associated with greater overall 28-day mortality compared to a low-to-moderate dose. NLR, a marker for individual immune response, varied between treatment groups.