. 2018 Feb 27;21(Suppl Suppl 1):e25043. doi: 10.1002/jia2.25043

Table 4.

Controlled direct effects of mediator interventions on physical intimate partner violence using parametric G‐computation formula

Mediator intervention	Risk (%) (T = treatment C = control)	Risk ratio (95% CI)	No mediator intervention
No sexual debut	T = 14.8	0.57 (0.48–0.65)	Risk (%) T = 18.5 C = 27.8 (RR 0.66)
No sexual debut	C = 26.2	0.57 (0.48–0.65)
No sexual partner (past 12 months)	T = 14.8	0.53 (0.46–0.60)
No sexual partner (past 12 months)	C = 27.9	0.53 (0.46–0.60)
Reduce sexual partners by one (past 12 months)	T = 16.2	0.59 (0.53–0.66)
Reduce sexual partners by one (past 12 months)	C = 27.4	0.59 (0.53–0.66)

Adjusted for age and baseline level of mediator. Binary outcomes estimated using a GEE log‐linear model and robust variance. Number of sexual partners estimated using a linear regression model with standard errors clustered at the individual level. Effects estimated from Monte Carlo sample (N = 5,000) with 95% CIs calculated from standard deviations of point estimates.