Gformula: G-formula Analysis

In adayim/causalMed: Causal Mediation Analysis in Survival settings

Description

Mediation analysis for time varying mediator, estimation based on g-formula. Output contains total effect, #' natrual direct effect and natural indirect effect for mediation or regular g-formula. data.frame will be returned.

Usage

Gformula(
  data,
  id.var,
  base.vars,
  exposure,
  outcome,
  time.var,
  models,
  intervention = NULL,
  init.recode = NULL,
  in.recode = NULL,
  out.recode = NULL,
  mc.sample = 10000,
  verbose = TRUE
)

Arguments

data	Data set to be sued
id.var	ID variable per subject.
base.vars	A vector of time fixed baseline variables.
exposure	Intervention/Exposure variable
outcome	Name of the outcome variable.
time.var	Time variable.
models	A list of models for the G-formula, including exposure model, covariate model (if any), mediator model (if any), outcome model or censoring model (if any). See details in spec_model.
intervention	A named list with a value of intervention on exposure. if kept as NULL (default), the natrual intervention cousre will be calculated. eg: list(natural = NULL, always = 1, never = 0)
init.recode	optional, recoding of variables done at the begaining of the Monte Carlo loop. Needed for operations initalize baseline variables. This is executed at begaining of the Monte Carlo g-formula, excuted only once at time 0.
in.recode	optional, On the fly recoding of variables done before the Monte Carlo loop starts. Needed to do any kind of functional forms for entry times. This is executed at each start of the Monte Carlo g-formula time steps
out.recode	optional, On the fly recoding of variables done at the end of the Monte Carlo loop. Needed for operations like counting the number of days with a treatment or creating lagged variables. This is executed at each end of the Monte Carlo g-formula time steps. @param is.survival Is the data survival data, defalt is FALSE.
mc.sample	Sample size of Monte Carlo simulation.
verbose	Print intervention information during calculation.

Note

Not that final outcome must be the same for in all rows per subject. If the dataset is survival settings, records after the interested outcome must be deleted. The funciton it self do some data manupilation internally. Please prepare the data as longitudinal format.

References

Lin, S. H., Young, J. G., Logan, R., & VanderWeele, T. J. (2017). Mediation analysis for a survival outcome with time‐varying exposures, mediators, and confounders. Statistics in medicine, 36(26), 4153-4166. DOI:10.1002/sim.7426 Zheng, W., & van der Laan, M. (2017). Longitudinal mediation analysis with time-varying mediators and exposures, with application to survival outcomes. Journal of causal inference, 5(2). DOI:10.1515/jci-2016-0006

TODO: weights, time varying intervention

adayim/causalMed documentation built on June 2, 2020, 4:11 p.m.