Factors associated with COVID-19 severe illness leading to hospitalization in Bangladesh: a matched case-control study

A retrospective matched case-control study included 250 patients hospitalized with severe COVID-19 and 250 age- and sex-matched controls who did not require hospitalization. Data on 48 variables were collected from both the cases and controls using a structured questionnaire via retrospective telephone interviews. Univariable conditional logistic regression was initially performed to identify variables associated with severe illness. Variables with p ≤ 0.10 were included in the multivariable model to identify factors independently associated with severe illness. Whole-genome sequencing (WGS) was performed as an exploratory analysis on virus samples from five purposively selected case-control pairs to assess whether unique viral mutations were associated with severe outcomes. Having multiple (≥2) comorbidities (adjusted odds ratio = 2.5, 95% confidence interval: 1.4–4.4, p = 0.001) and attempting oxygen supplementation at home during the illness (adjusted odds ratio = 6.3, 95% confidence interval: 2.7–15.1, p < 0.001) were associated with higher odds of severe COVID-19 leading to hospitalization. In contrast, indoor physical exercise during the illness was associated with lower odds of severe illness (adjusted odds ratio = 0.3, 95% confidence interval: 0.2–0.5, p < 0.001). Both exposures occurred after symptom onset; these associations may reflect delayed hospital care (home oxygen) and pre-illness physical activity or milder disease (exercise), and reverse causation cannot be excluded. In the exploratory analysis of ten purposively selected genomes, no consistent mutations were associated with severe illness; the small sample cannot exclude a viral contribution to severity.