Duval & Tweedie Trim-and-Fill Publication Bias Adjustment — trim_and_fill • pharmhand

Performs the Duval & Tweedie trim-and-fill method to estimate the number of missing studies and adjust the pooled effect for publication bias.

Usage

trim_and_fill(
  meta_result,
  side = c("auto", "left", "right"),
  estimator = c("L0", "R0", "Q0"),
  maxiter = 100
)

Arguments

meta_result: A MetaResult object from meta_analysis()
side: Character. Side where excess studies are trimmed: "left", "right", or "auto". Missing studies are imputed on opposite side. Default: "auto"
estimator: Character. Method to estimate missing studies: "L0", "R0", "Q0". Default: "L0"
maxiter: Integer. Maximum iterations. Default: 100

Value

A list with components:

original: List with original estimate, confidence interval, and k
adjusted: List with adjusted estimate, confidence interval, and k
n_imputed: Estimated number of missing studies
side: Side where studies were imputed
imputed_studies: Data frame with imputed effects
estimator: Estimator method used
effect_measure: Effect measure used
interpretation: Text interpretation of the adjustment

Examples

# Trim-and-fill for publication bias adjustment
yi <- c(-0.5, -0.4, -0.3, -0.1, 0.0, 0.5, 0.6)
sei <- c(0.1, 0.12, 0.15, 0.18, 0.2, 0.1, 0.12)
meta_res <- meta_analysis(yi = yi, sei = sei, effect_measure = "md")
adjusted <- trim_and_fill(meta_res)
adjusted$n_imputed
#> [1] 0
adjusted$interpretation
#> [1] "No missing studies detected"