Skip to main content
. 2013 Apr 18;3(4):e002157. doi: 10.1136/bmjopen-2012-002157

Table 5.

Proportion of mediation of smoking to mortality association explained by each complication in unadjusted and adjusted models

Outcome Mediator Proportion mediation unadjusted for covariates (%) Proportion mediation adjusted for covariates (%)
6 months mortality Surgical site infection 22.2 2.0
Cardiovascular complications 12.2 3.8
Pulmonary complications 88.9 21.6
Composite outcome 86.4 15.7
1 year mortality Surgical site infection 12.6 1.7
Cardiovascular complications 6.5 3.0
Pulmonary complications 42.7 16.2
Composite outcome 40.8 11.3

Because the proportion mediated is ratio statistic, its estimated value is sensitive to variation in point estimates of the regression coefficients from which it is derived; it should therefore be interpreted with caution. Coefficient values (magnitude and significance) should be the main proponent in assessing mediation.