. 2023 Jan 12;60(3):103276. doi: 10.1016/j.ipm.2023.103276

Table 4.

Mediation Analysis of Causal Effects that illustrate the different paths of the influence of sex on COVID-19 severity. All effects except Direct Effect indicate a severity bias for men (positive values). The Direct Effect is close to zero, because we assume through the causal graph used as prior model of the world that all the influence of sex/gender on COVID-19 severity is explained by the mediating variables (either BioVar or Lifestyle variables) The effect caused by BioVar mediating variable is higher than the effect caused by the Lifestyle mediating variable. The last two columns of the table indicate lower and upper bounds for confidence intervals for the estimated effect values.

Variable	Estimated effect	Standard error	CI lower 95%	CI upper 95%
Natural Direct Effect:
$S e x - > C O V I D S e v e r i t y$	−0.015	0.035	−0.061	0.048
Path-Specific (Indirect) Effect :
$S e x - > B i o V a r - > C O V I D S e v e r i t y$	0.521	0.039	0.498	0.609
Path-Specific (Indirect) Effect:
$S e x - > L i f e s t y l e - > C O V I D S e v e r i t y$	0.148	0.031	0.084	0.168
Total Effect:
Sex $- > C O V I D S e v e r i t y$	0.654	0.008	0.644	0.668
Total Variation:
$S e x - > C O V I D S e v e r i t y$	0.7081	–	–	–
Confounding Effect:
$S e x < - G e n d e r R o l e s - > C O V I D S e v e r i t y$	0.0541	–	–	–