Implementation of extended formulas when there are effect measure modifiers

In regmedint: Regression-Based Causal Mediation Analysis with Interaction and Effect Modification Terms

## knitr configuration: http://yihui.name/knitr/options#chunk_options
library(knitr)
showMessage <- FALSE
showWarning <- TRUE
set_alias(w = "fig.width", h = "fig.height", res = "results")
opts_chunk$set(comment = "##", error= TRUE, warning = showWarning, message = showMessage,
               tidy = FALSE, cache = FALSE, echo = TRUE,
               fig.width = 7, fig.height = 7,
               fig.path = "man/figures")

In this document, we demonstrate including effect measure modification (EMM) terms in the mediator or the outcome models. The dataset used in this document is still vv2015.

library(regmedint)
library(tidyverse)
## Prepare dataset
data(vv2015)

No EMM by covariates

In the first model fit, we do not include any EMM term.

regmedint_obj1 <- regmedint(data = vv2015,
                            ## Variables
                            yvar = "y",
                            avar = "x",
                            mvar = "m",
                            cvar = c("c"),
                            eventvar = "event",
                            ## Values at which effects are evaluated
                            a0 = 0,
                            a1 = 1,
                            m_cde = 1,
                            c_cond = 3,
                            ## Model types
                            mreg = "logistic",
                            yreg = "survAFT_weibull",
                            ## Additional specification
                            interaction = TRUE,
                            casecontrol = FALSE)
summary(regmedint_obj1)

EMM by covariates

There is $A\times C$ term in mediator model

Now suppose the covariate $C$ modifies the treatment effect on the mediator. We add emm_ac_mreg = c("c") in regmedint(). Although there is only one covariate in our dataset, emm_ac_mreg can take a vector of multiple covariates. Please note that the covariates in emm_ac_mreg should be a subset of the covariates specified in cvar, i.e. if a covariate is an effect measure modifier included in emm_ac_mreg, it must be included in cvar, otherwise an error message will be printed.

regmedint_obj2 <- regmedint(data = vv2015,
                            ## Variables
                            yvar = "y",
                            avar = "x",
                            mvar = "m",
                            cvar = c("c"),
                            emm_ac_mreg = c("c"),
                            emm_ac_yreg = NULL,
                            emm_mc_yreg = NULL,
                            eventvar = "event",
                            ## Values at which effects are evaluated
                            a0 = 0,
                            a1 = 1,
                            m_cde = 1,
                            c_cond = 3,
                            ## Model types
                            mreg = "logistic",
                            yreg = "survAFT_weibull",
                            ## Additional specification
                            interaction = TRUE,
                            casecontrol = FALSE)
summary(regmedint_obj2)

There is $A\times C$ term in both mediator and outcome models

Now suppose in addition to the EMM on mediator, the covariate $C$ also modifies the treatment effect on the outcome We add emm_ac_yreg = c("c") in regmedint(). Please note that the covariates in emm_ac_yreg should be a subset of the covariates specified in cvar, i.e. if a covariate is an effect measure modifier included in emm_ac_yreg, it must be included in cvar, otherwise an error message will be printed.

regmedint_obj3 <- regmedint(data = vv2015,
                            ## Variables
                            yvar = "y",
                            avar = "x",
                            mvar = "m",
                            cvar = c("c"),
                            emm_ac_mreg = c("c"),
                            emm_ac_yreg = c("c"),
                            emm_mc_yreg = NULL,
                            eventvar = "event",
                            ## Values at which effects are evaluated
                            a0 = 0,
                            a1 = 1,
                            m_cde = 1,
                            c_cond = 3,
                            ## Model types
                            mreg = "logistic",
                            yreg = "survAFT_weibull",
                            ## Additional specification
                            interaction = TRUE,
                            casecontrol = FALSE)
summary(regmedint_obj3)

There are $A\times C$ term in both mediator and outcome models, and $M\times C$ term in outcome model

Now suppose in addition to the EMM of treatment effect, the covariate $C$ also modifies the mediator effect on the outcome. We add emm_mc_yreg = c("c") in regmedint(). Please note that the covariates in emm_mc_yreg should be a subset of the covariates specified in cvar, i.e. if a covariate is an effect measure modifier included in emm_mc_yreg, it must be included in cvar, otherwise an error message will be printed.

regmedint_obj4 <- regmedint(data = vv2015,
                            ## Variables
                            yvar = "y",
                            avar = "x",
                            mvar = "m",
                            cvar = c("c"),
                            emm_ac_mreg = c("c"),
                            emm_ac_yreg = c("c"),
                            emm_mc_yreg = c("c"),
                            eventvar = "event",
                            ## Values at which effects are evaluated
                            a0 = 0,
                            a1 = 1,
                            m_cde = 1,
                            c_cond = 3,
                            ## Model types
                            mreg = "logistic",
                            yreg = "survAFT_weibull",
                            ## Additional specification
                            interaction = TRUE,
                            casecontrol = FALSE)
summary(regmedint_obj4)

regmedint documentation built on April 7, 2022, 1:17 a.m.