Table 3

Total direct effect of smoking on outcomes,estimates using a directed acyclic graph (‘causal diagram’)

WHO ordinal scale, n (%)Ever-smoker
(n=126)
Non-smoker
(n=321)
Adjusted estimate
(adjusted OR, 95% CI, P value) *
1.34 (0.85 to 2.12, p=0.210)
No oxygen21 (16.7)81 (25.2)
Oxygen mask or nasal canula86 (68.3)191 (59.5)0.93 (0.86 to 1.01, p=0.103)
Non-invasive mechanical ventilation or high-flow nasal canula9 (7.1)10 (3.1)1.07 (0.95 to 1.18, p=0.240)
Invasive mechanical ventilation2 (1.6)6 (1.9)1.04 (1.00 to 1.09, p=0.080)
Death8 (6.3)33 (10.3)1.00 (0.97 to 1.03, p=0.907)
ICU admission, n (%)22 (17.5)53 (16.5)0.97 (0.91 to 1.03, p=0.321)
ICU length of stay in days, median (95% CI)10 (6.0 to 21.0)8.8 (5.0 to 10.0)2.06 (0.76 to 5.59, p=0.158)
Hospital length of stay in days, median (95% CI)8.6 (7 to 10)6.9 (6.0 to 7.0)1.52 (1.05 to 2.20, p=0.027)
  • *Adjustment set (age, gender, body mass index and COVID-19 medications). Ordinal regression model was fitted. A logistic regression model was fitted for individual endpoints.

  • ICU, intensive care unit.