In fmmattioni/metabolic: Datasets and Functions for Reproducing Meta-Analyses

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  warning = FALSE,
  fig.align = "center"
)
options(width = 999)

# from https://github.com/r-lib/crayon/issues/24#issuecomment-581068792
# crayon needs to be explicitly activated in Rmd
options(crayon.enabled = TRUE)
# Hooks needs to be set to deal with outputs
# thanks to fansi logic
old_hooks <- fansi::set_knit_hooks(knitr::knit_hooks,
                                   which = c("output", "message", "error"))

The {metabolic} package provides you all the tools necessary to reproduce the meta-analysis published in Medicine & Science in Sports & Exercise.

You can easily read the paper in the journal website through the function metabolic::read_paper(). It will take you to the online published paper in your default browser.

library(metabolic)

read_paper()

```{block, type = 'rmdinfo'} We also provide you all the data used to perform the meta-analysis and the GOfER (Graphical Overview for Evidence Reviews) diagram. In case you are not familiar with R, or just would like to download the data, you can do so in the metabolic_meta and in the metabolic_gofer vignettes.

## Produce an HTML report for a given clinical endpoint

The easiest way of reproducing the meta-analysis for a given clinical endpoint is using the `metabolic::build_report()`function, which will build an **HTML report** with all the results, including **R output** and **plots**. Try it out! Make sure you also choose a path to save the report to!

```r
library(metabolic)

build_report(endpoint = "VO2max", path = "~/Documents/VO2max_report")

```{block, type = 'rmdinfo'} In case you want to reproduce the analysis on your own, we provided you the necessary tools for that! Here we will walk you through all the functionalities. In this example, we will be reproducing a meta-analysis on VO2max.

## Perform the meta-analysis

```r
library(metabolic)

results <- perform_meta(endpoint = "VO2max")

results

This function is going to perform the overall meta-analysis, followed by a sensitivity analysis where the meta-analysis in run again omitting one study at a time. This is done to ensure that the meta-analysis results are robust and is not being influenced by a single study.

In case the sensitivity analysis detects that there is a single study influencing the results, the study identifier (Author Year) is printed, and the meta-analysis is re-done without that study. Importantly, this study is excluded from the overall meta-analysis, but is still considered in subgroup analyses.

```{block, type = 'rmdinfo'} As an example, here is the output from the HbA1c meta-analysis, where there sensitivity analysis recognizes a single-study influence on the results:

```r
perform_meta(endpoint = "HbA1c")

Small-study effects

Small-study effects (the phenomenon which smaller studies present different - often larger - treatment effects than bigger sample studies) may be visualized through contour-enhanced funnel plots, as well as through radial plots.

results$meta_analysis$Overall %>% 
  plot_small_study_effects()

Sensitivity analysis

You may also access the results from the sensitivity analysis and produce a forest plot:

results$sensitivity_analysis$Overall

results$sensitivity_analysis$Overall %>% 
  plot_metabolic()

Overall meta-analysis

The overall meta-analysis results can be easily accessed through .$meta_analysis$Overall:

results$meta_analysis$Overall

And the forest plot is generated through the generic function metabolic::plot_metabolic():

results$meta_analysis$Overall %>% 
  plot_metabolic()

Subgroups meta-analysis

Overview

The overview of the subgroups meta-analysis results can be easily generated through the function metabolic::perform_bind()

results_bind <- perform_bind(results$meta_analysis)

results_bind

And the forest plot can be generated through the generic function metabolic::plot_metabolic():

results_bind %>% 
  plot_metabolic()

Subgroups

Each subgroup analysis will have its own meta-analysis and meta-regression. Here is an example for the Population and Age subgroups:

Population

results$meta_analysis$Population

results$meta_analysis$Population %>% 
  plot_metabolic()

results$meta_regression$Population

results$meta_regression$Population %>% 
  plot_metabolic()

Age

results$meta_analysis$Age

results$meta_analysis$Age %>% 
  plot_metabolic()

results$meta_regression$Age

results$meta_regression$Age %>% 
  plot_metabolic()

fmmattioni/metabolic documentation built on Oct. 17, 2023, 2:41 a.m.