Skip to main content

Table 3 The interaction effect estimates between tinnitus and several additional variables. Confounding was adjusted for via both regression adjustment and IPTW, firstly when no moderator-confounder interactions were accounted for in the adjustment model and secondly when all possible moderator-confounder interactions were accounted for in the adjustment model

From: The impact of moderator by confounder interactions in the assessment of treatment effect modification: a simulation study

  Regression adjustment IPTW
No \({\varvec{M}}\times {\varvec{X}}\) interactions All \({\varvec{M}}\times {\varvec{X}}\) interactions No \({\varvec{M}}\times {\varvec{X}}\) interactions All \({\varvec{M}}\times {\varvec{X}}\) interactions
Estimate
(95% CI)
Estimate
(95% CI)
Estimate
(95% CI)
Estimate
(95% CI)
Gender, female 2.56 (-0.80, 5.92) 2.86 (-0.52, 6.23) 2.60 (-0.84, 6.04) 2.37 (-1.07, 5.82)
White ethnicity -0.92 (-10.13, 8.28) -2.84 (-12.10, 6.43) -2.84 (-12.97, 7.29) -4.24 (-15.06, 6.57)
Current smoking -3.72 (-7.95, 0.50) -3.93 (-8.17, 0.31) -3.62 (-7.86, 0.62) -3.48 (-7.70, 0.74)