some basic work on developing the package

* Rdpack added
* @param and text added to function
* testthat scaffolding

DESCRIPTION

@@ -1,19 +1,26 @@
 Package: mimR
 Title: Multiple Imputation Marginalisation
 Version: 0.0.0.9000
-Authors@R: 
+Authors@R: c(
     person("Nathan", "Green", , "[email protected]",
                   role = c("aut", "cre"),
-                  comment = c(ORCID = "0000-0003-2745-1736"))
-Description: Multiple Imputation Marginalisation.
+                  comment = c(ORCID = "0000-0003-2745-1736")),
+    person("Antonio", "Remiro-Azocar", email = "[email protected]",
+                  role = c("aut"),
+                  comment = c(ORCID = "0000-0002-7263-4251")))
+Description: Multiple Imputation Marginalisation with several different computation approaches.
+    See Remiro‐Azócar A, Heath A, Baio G (2022). “Parametric G‐computation for compatibl
+    indirect treatment comparisons with limited individual patient data.”
+    Res. Synth. Methods, 1–31. ISSN 1759-2879, doi:10.1002/jrsm.1565, 2108.12208
 License: GPL (>= 3)
 Encoding: UTF-8
 Roxygen: list(markdown = TRUE)
 RoxygenNote: 7.2.3.9000
 Imports:
     boot,
     copula,
-    rstanarm
+    rstanarm,
+    Rdpack (>= 0.7)
 Suggests:
     dplyr,
     ggplot2,
@@ -26,3 +33,4 @@ Suggests:
     testthat (>= 3.0.0)
 VignetteBuilder: knitr
 Config/testthat/edition: 3
+RdMacros: Rdpack

NAMESPACE

@@ -24,3 +24,4 @@ export(strategy_maic)
 export(strategy_stc)
 export(trial_treatment_effect)
 export(trial_variance)
+importFrom(Rdpack,reprompt)

R/IPD_stats.R

@@ -2,13 +2,20 @@
 # create class for each approach
 
 #' @rdname strategy
-#' strategy_maic
+#' Matching-adjusted indirect comparison strategy Used to compare marginal
+#' treatment effects where there are cross-trial differences in effect modifiers
+#' and limited patient-level data.
 #'
-#' @param formula 
+#' The default formula is
+#' \deqn{
+#'  y = X3 + X4 + trt*X1 + trt*X2
+#' }
+#' 
+#' @param formula Linear regression formula object 
 #' @param R 
-#' @param ald 
+#' @param ald Aggregate-level data 
 #'
-#' @return
+#' @return `maic` class object
 #' @export
 #'
 strategy_maic <- function(formula = as.formula("y ~ X3 + X4 + trt*X1 + trt*X2"),
@@ -21,11 +28,17 @@ strategy_maic <- function(formula = as.formula("y ~ X3 + X4 + trt*X1 + trt*X2"),
 }
 
 #' @rdname strategy
-#' strategy_stc
-#'
-#' @param formula 
+#' Simulated treatment comparison strategy
+#' outcome regression-based method which targets a conditional treatment effect.
+#' 
+#' The default formula is
+#' \deqn{
+#'  y = X3 + X4 + trt*I(X1 - mean(X1)) + trt*I(X2 - mean(X2))
+#' }
+#' 
+#' @param formula Linear regression formula object
 #'
-#' @return
+#' @return `stc` class object
 #' @export
 # 
 strategy_stc <- function(formula =
@@ -39,12 +52,17 @@ strategy_stc <- function(formula =
 }
 
 #' @rdname strategy
-#' strategy_gcomp_ml
+#' G-computation maximum likelihood strategy
 #'
-#' @param formula 
+#' @param formula Linear regression formula object
 #' @param R 
 #'
-#' @return
+# The default formula is
+#' \deqn{
+#'  y = X3 + X4 + trt*X1 + trt*X2)
+#' }
+#' 
+#' @return `gcomp_ml` class object
 #' @export
 #'
 strategy_gcomp_ml <- function(formula =
@@ -57,11 +75,16 @@ strategy_gcomp_ml <- function(formula =
 }
 
 #' @rdname strategy
-#' strategy_gcomp_stan
+#' G-computation Bayesian strategy
 #'
-#' @param formula 
+#' The default formula is
+#' \deqn{
+#'  y = X3 + X4 + trt*X1 + trt*X2)
+#' }
+#' 
+#' @param formula Linear regression formula object
 #'
-#' @return
+#' @return `gcomp_stan` class object
 #' @export
 #'
 strategy_gcomp_stan <- function(formula =
@@ -73,9 +96,11 @@ strategy_gcomp_stan <- function(formula =
 }
 
 #' @name strategy
-#' New_strategy
+#' New strategy objects
+#' 
+#' Create class for each approach
 #'
-#' @param strategy 
+#' @param strategy Class name
 #' @param ... Additional arguments
 #'
 #' @return
@@ -86,12 +111,23 @@ new_strategy <- function(strategy, ...) {
 }
 
 
-#' Main wrapper for hat_Delta_stats
+#' Calculate the difference between treatments using all evidence
+#' 
+#' This is the main wrapper for `hat_Delta_stats()`.
 #' 
-#' @param AC.IPD 
-#' @param BC.ALD 
-#' @param strategy 
+#' \insertCite{RemiroAzocar2022}{mimR}
+#' 
+#' @param AC.IPD Individual-level patient data. Suppose between studies A and C.
+#' @param BC.ALD Aggregate-level data. Suppose between studies B and C. 
+#' @param strategy Computation strategy function. These can be
+#'    `strategy_maic()`, `strategy_stc`, `strategy_gcomp_ml` and  `strategy_gcomp_stan`
 #' @param ... Additional arguments
+#' @return List of statistics as a `mimR` class object
+#' @importFrom Rdpack reprompt
+#' 
+#' @references
+#' \insertRef{RemiroAzocar2022}{mimR}
+#' 
 #' @export
 #' 
 hat_Delta_stats <- function(AC.IPD, BC.ALD, strategy, ...) {
@@ -126,7 +162,13 @@ hat_Delta_stats <- function(AC.IPD, BC.ALD, strategy, ...) {
 
 #' @name IPD_stats
 #' Individual level data statistics
-#' @return mean, variance
+#' 
+#' @param strategy A list corresponding to different approaches
+#' @template args-ipd
+#' @template args-ald
+#' @param ... Additional arguments
+#' 
+#' @return Mean and variance
 #' @export
 #' 
 IPD_stats <- function(strategy, ipd, ald, ...)
@@ -141,14 +183,15 @@ IPD_stats.default <- function() {
 
 
 #' @rdname IPD_stats
-#' IPD_stats.maic
+#' Matching-adjusted indirect comparison statistics
 #' 
 #' marginal A vs C treatment effect estimates
 #' using bootstrapping
 #'
-#' @param strategy 
+#' @param strategy A list corresponding to different approaches
 #' @template args-ipd
 #' @template args-ald
+#' @return Mean and variance
 #'
 #' @export
 #' 
@@ -171,11 +214,16 @@ IPD_stats.maic <- function(strategy,
 
 
 #' @rdname IPD_stats
-#' IPD_stats.stc
+#' Simulated treatment comparison statistics 
+#' 
+#' IPD from the AC trial are used to fit a regression model describing the
+#' observed outcomes `y` in terms of the relevant baseline characteristics `x` and
+#' the treatment variable `z`.
 #' 
-#' @param strategy 
+#' @param strategy A list corresponding to different approaches
 #' @template args-ipd
 #' @template args-ald
+#' @return Mean and variance
 #' @export
 #' 
 IPD_stats.stc <- function(strategy,
@@ -193,13 +241,13 @@ IPD_stats.stc <- function(strategy,
 
 
 #' @rdname IPD_stats
-#' IPD_stats.gcomp_ml
+#' G-computation maximum likelihood statistics
 #'
-#' @param strategy 
+#' @param strategy A list corresponding to different approaches
 #' @template args-ipd
 #' @template args-ald
 #'
-#' @return
+#' @return Mean and variance
 #' @export
 #'
 IPD_stats.gcomp_ml <- function(strategy,
@@ -217,13 +265,13 @@ IPD_stats.gcomp_ml <- function(strategy,
 
 
 #' @rdname IPD_stats
-#' IPD_stats.gcomp_stan
+#' G-computation Bayesian statistics
 #'
-#' @param strategy 
+#' @param strategy A list corresponding to different approaches
 #' @template args-ipd
 #' @template args-ald
 #'
-#' @return
+#' @return Mean and variance
 #' @export
 #'
 IPD_stats.gcomp_stan <- function(strategy,

R/gcomp_stan.R

@@ -1,11 +1,11 @@
 
 #' G-computation using Stan
 #'
-#' @param formula 
+#' @param formula Linear regression formula object 
 #' @template args-ipd
 #' @template args-ald
 #'
-#' @return
+#' @return A list of `y.star.A` and `y.star.C` posterior predictions
 #' @export
 #'
 gcomp_stan <- function(formula = as.formula("y ~ X3 + X4 + trt*X1 + trt*X2"),

inst/REFERENCES.bib

@@ -0,0 +1,10 @@
+@article{RemiroAzocar2022,
+author = {Remiro‐Az{\'{o}}car, Antonio and Heath, Anna and Baio, Gianluca},
+doi = {10.1002/jrsm.1565},
+eprint = {2108.12208},
+issn = {1759-2879},
+journal = {Res. Synth. Methods},
+pages = {1--31},
+title = {{Parametric G‐computation for compatible indirect treatment comparisons with limited individual patient data}},
+year = {2022}
+}

tests/testthat/test-hat_Delta_stats.R

@@ -0,0 +1,3 @@
+test_that("multiplication works", {
+  expect_equal(2 * 2, 4)
+})

vignettes/Example.Rmd → vignettes/Basic_example.Rmd

@@ -19,6 +19,10 @@ knitr::opts_chunk$set(
 
 ## Introduction
 
+Population adjustment methods such as matching-adjusted indirect comparison (MAIC) are increasingly used to compare marginal treatment effects when there are cross-trial differences in effect modifiers and limited patient-level data. MAIC is based on propensity score weighting, which is sensitive to poor covariate over- lap and cannot extrapolate beyond the observed covariate space. Current out- come regression-based alternatives can extrapolate but target a conditional treatment effect that is incompatible in the indirect comparison. When adjusting for covariates, one must integrate or average the conditional estimate over the rel- evant population to recover a compatible marginal treatment effect. We propose a marginalization method based on parametric G-computation that can be easily applied where the outcome regression is a generalized linear model or a Cox model. The approach views the covariate adjustment regression as a nuisance model and separates its estimation from the evaluation of the marginal treatment effect of interest. The method can accommodate a Bayesian statistical framework, which naturally integrates the analysis into a probabilistic framework. A simula- tion study provides proof-of-principle and benchmarks the method's performance against MAIC and the conventional outcome regression. Parametric G- computation achieves more precise and more accurate estimates than MAIC, par- ticularly when covariate overlap is poor, and yields unbiased marginal treatment effect estimates under no failures of assumptions. Furthermore, the marginalized regression-adjusted estimates provide greater precision and accuracy than the conditional estimates produced by the conventional outcome regression, which are
+
+## Example
+
 ```{r setup, warning=FALSE, message=FALSE}
 library(boot)      # non-parametric bootstrap in MAIC and ML G-computation
 library(copula)    # simulating BC covariates from Gaussian copula