apply_contrast {beeca}

Apply contrast to calculate marginal estimate of treatment effect and corresponding standard error

Description

Calculates the marginal estimate of treatment effect and its corresponding standard error based on a fitted GLM object using specified contrast (summary measure) methods

Usage

apply_contrast(
  object,
  contrast = c("diff", "rr", "or", "logrr", "logor"),
  reference
)

Arguments

object

a fitted glm object augmented with counterfactual.predictions, counterfactual.means and robust_varcov.

contrast

a string specifying the type of contrast to apply. Accepted values are "diff" (risk difference), "rr" (risk ratio), "or" (odds ratio), "logrr" (log risk ratio), "logor" (log odds ratio). Note: log-transformed ratios (logrr and logor) work better compared to rr and or when computing confidence intervals using normal approximation. The choice of contrast affects how treatment effects are calculated and interpreted. Default is diff.

reference

a string indicating which treatment group should be considered as the reference level. Accepted values are one of the levels in the treatment variable. Default to the first level used in the glm object.

This parameter influences the calculation of treatment effects relative to the chosen reference group.

Details

The apply_constrast() functions computes the summary measure between two arms based on the estimated marginal effect and its variance-covariance matrix using the Delta method.

Note: Ensure that the glm object has been adequately prepared with average_predictions() and estimate_varcov() before applying apply_contrast(). Failure to do so may result in errors indicating missing components.

Value

An updated glm object with two additional components appended: marginal_est (marginal estimate of the treatment effect) and marginal_se (standard error of the marginal estimate). These appended component provide crucial information for interpreting the treatment effect using the specified contrast method.

See Also

get_marginal_effect() for estimating marginal effects directly from an original glm object

Examples

trial01$trtp <- factor(trial01$trtp)
fit1 <- glm(aval ~ trtp + bl_cov, family = "binomial", data = trial01) |>
  predict_counterfactuals(trt = "trtp") |>
  average_predictions() |>
  estimate_varcov(method = "Ye") |>
  apply_contrast("diff", reference = "0")

# Assuming `trial01` is a dataset with treatment (`trtp`)
# and baseline covariate (`bl_cov`)
trial01$trtp <- factor(trial01$trtp)
fit1 <- glm(aval ~ trtp + bl_cov, family = "binomial", data = trial01)

# Preprocess fit1 as required by apply_contrast
fit2 <- fit1 |>
  predict_counterfactuals(trt = "trtp") |>
  average_predictions() |>
  estimate_varcov(method = "Ye")

# Apply contrast to calculate marginal estimates
fit3 <- apply_contrast(fit2, contrast = "diff", reference = "0")

fit3$marginal_est
fit3$marginal_se