Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/namakala/inappropriate-acid-suppressor-agent-use/llms.txt

Use this file to discover all available pages before exploring further.

meta-analysis.R provides four functions that implement the full statistical modeling pipeline. fitMetaprop() fits an overall proportion meta-analysis, fitSubMetaprop() extends it to subgroup analyses, fitMetareg() fits univariable or multivariable meta-regression, and applyCopas() applies the Copas selection model to adjust for publication bias.

​
fitMetaprop()

fitMetaprop <- function(tbl, ...) {
  #' Meta-Analysis for Proportion
  #'
  #' Fit meta-analyis model for proportion data with REML estimator,
  #' Freeman-Tukey summary measure, and Hartung-Knapp CI calculation.
  #'
  #' @param tbl A data frame object containing extracted information from the
  #' selected articles
  #' @return Metaprop object
  require("meta")

  mod <- meta::metaprop(
    data             = tbl,
    event            = Inappropriate_indication,
    n                = Sample_size,
    studlab          = Author,
    sm               = "PFT",
    method.tau       = "REML",
    method.random.ci = "HK",
    common           = FALSE,
    random           = TRUE,
    prediction       = TRUE,
    ...
  )

  return(mod)
}

​
Parameters

​
tbl
data.frame
required
Cleaned data frame from clean(). Must contain Inappropriate_indication (event count), Sample_size (total n), and Author (study label).
​
...
any
Additional arguments passed to meta::metaprop(). Use these to override model settings such as subgroup or keepdata.

​
Returns

A metaprop object from the meta package. Key slots include TE.random, lower.random, upper.random, I2, tau2, H, Q, k, n, and event.

​
Model settings

SettingValueDescription
sm"PFT"Freeman-Tukey double arcsine transformation
method.tau"REML"Restricted maximum likelihood for heterogeneity
method.random.ci"HK"Hartung-Knapp confidence intervals
commonFALSECommon-effect model not fitted
randomTRUERandom-effects model fitted
predictionTRUEPrediction interval included

​
Usage

targets::tar_load(tbl_clean)
mod <- fitMetaprop(tbl_clean)
summary(mod)

​
fitSubMetaprop()

fitSubMetaprop <- function(tbl, varname, ...) {
  #' Subgroup Meta-Analysis
  #'
  #' Perform a subgroup meta-analysis grouped by the given `varname`.
  #'
  #' @param tbl A data frame object
  #' @param varname Variable name to group the analysis
  #'
  #' @return Metaprop object

  idx     <- !is.na(tbl[[rlang::as_name(varname)]])
  sub_tbl <- tbl |> dplyr::filter(idx)

  mod <- sub_tbl %>% fitMetaprop(
    subgroup = dplyr::pull(., {{ varname }}), keepdata = TRUE
  )

  return(mod)
}

​
Parameters

​
tbl
data.frame
required
Cleaned data frame from clean().
​
varname
symbol
required
A rlang symbol identifying the grouping variable. Valid columns in the pipeline are Age, Continent, Setting, JBI_Classification, and use_guideline. Pass as rlang::sym("Continent").
​
...
any
Additional arguments passed to fitMetaprop().

​
Returns

A metaprop object with subgroup structure. The subgroup slot contains the grouping variable values; TE.random.w, lower.random.w, and upper.random.w hold the per-subgroup estimates.

​
How it works

The function first drops rows where varname is NA using rlang::as_name() to convert the symbol to a string for subsetting. It then calls fitMetaprop() with the subgroup argument set to the pulled column and keepdata = TRUE so the original data is retained in the model object.

​
Usage

targets::tar_load(tbl_clean)
mod_sub <- fitSubMetaprop(tbl_clean, varname = rlang::sym("Continent"))
In _targets.R: 
tar_target(mod_subgroup, fitSubMetaprop(tbl_clean, varname = rlang::sym(colname)))

​
fitMetareg()

fitMetareg <- function(tbl, varname, ...) {
  #' Meta-Regression
  #'
  #' Fit a meta-regression model to a given data frame and set of variable
  #' names. Variable names could be singular, which then will be fitted as a
  #' univariate meta-regression model.
  #'
  #' @param tbl A data frame object
  #' @param varname One or multiple column names
  #' 
  #' @return A `rma` object
  require("metafor")
  require("meta")

  sub_tbl <- tbl |>
    subset(
      subset = var_logit_prevalence != Inf,
      select = c("logit_prevalence", "var_logit_prevalence", varname)
    ) |>
    na.omit()

  is_single <- length(varname) == 1

  if (is_single) {
    form <- paste("~", varname)
  } else {
    form <- paste("~", paste(varname, collapse = " + "))
  }

  form %<>% as.formula()

  mod <- metafor::rma(
    yi     = logit_prevalence,
    vi     = var_logit_prevalence,
    mods   = form,
    data   = sub_tbl,
    method = "REML",
    ...
  )

  return(mod)
}

​
Parameters

​
tbl
data.frame
required
Cleaned data frame from clean().
​
varname
character
required
A character vector of column names to use as moderators. A single element fits a univariable model; two or more elements fit a multivariable model with all terms joined by +.
​
...
any
Additional arguments passed to metafor::rma().

​
Returns

An rma object from the metafor package. Key slots include b (coefficients), ci.lb, ci.ub, pval, I2, tau2, QE, and QEp.

​
How it works

Before fitting, the function removes rows where var_logit_prevalence is Inf and drops any remaining NA rows via na.omit(). The formula is constructed dynamically: a single varname produces ~ varname; multiple names produce ~ var1 + var2 + .... The outcome is logit_prevalence and the weights are var_logit_prevalence, both fitted with REML.

​
Usage

targets::tar_load(tbl_clean)
mod_uni <- fitMetareg(tbl_clean, "Continent")

​
applyCopas()

applyCopas <- function(obj, varname = NULL, ...) {
  #' Copas Selection objel
  #'
  #' Perform a Copas selection objel analysis on a given meta-analysis objel.
  #' This analysis will adjust for the probability of publication bias.
  #'
  #' @param obj A meta-analysis model from the `meta` package, otherwise a data
  #' frame object.
  #' @param varname Variable name indicating which variable is used to subste
  #' the dataset.
  #'
  #' @return A Copas analysis object
  require("metasens")
  require("meta")

  tbl <- obj

  if (grepl(x = obj, "meta") |> any()) {
    mod_copas <- metasens::copas(obj, backtransf = TRUE, ...)
    return(mod_copas)
  } else if (is.null(varname)) {
    mod <- fitMetaprop(obj)
    mod_copas <- applyCopas(mod)
    return(mod_copas)
  }

  groups <- tbl[[varname]] |> unique() |> na.omit()
  sub_tbls <- lapply(groups, \(group) tbl %>% subset(.[[varname]] == group))
  sub_metas <- lapply(sub_tbls, \(sub_tbl) fitMetaprop(sub_tbl))
  mod_copas <- lapply(sub_metas, applyCopas)

  return(mod_copas)
}

​
Parameters

​
obj
meta object or data.frame
required
Either a fitted meta object (e.g., from fitMetaprop()) for direct Copas analysis, or a raw data frame when you want applyCopas() to handle fitting internally.
​
varname
character
default:"NULL"
Column name to use for subgroup iteration. When NULL, Copas analysis is applied to the overall model. When provided along with a data frame obj, the function splits the data by unique levels of varname and returns one Copas object per subgroup.
​
...
any
Additional arguments passed to metasens::copas().

​
Returns

A copas object when applied to a single model, or a list of copas objects (one per subgroup level) when varname is provided with a data frame.

​
How it dispatches

The function checks class(obj) for the string "meta" using grepl():
  1. obj is a meta object — calls metasens::copas(obj, backtransf = TRUE, ...) directly and returns the result.
  2. obj is a data frame and varname is NULL — calls fitMetaprop(obj) first, then recurses with the resulting model.
  3. obj is a data frame and varname is given — splits the data into per-group subsets, fits fitMetaprop() on each, and applies applyCopas() recursively to produce a list.

​
Usage

targets::tar_load(mod_prop)
mod_copas <- applyCopas(mod_prop)
summary(mod_copas)
plot(mod_copas)
In _targets.R: 
# Overall
tar_target(mod_copas_prop, applyCopas(mod_prop))

# Subgroup
tar_target(mod_copas_subgroup, applyCopas(tbl_clean, varname = colname))

Build docs developers (and LLMs) love

Get started for freeTalk to us