| Type: | Package |
| Title: | Meaningful Grouping of Studies in Meta-Analysis |
| Version: | 1.0.2 |
| Date: | 2025-08-31 |
| Author: | Ahmed Abdelmageed 
[cre, aut] |
| Maintainer: | Ahmed Abdelmageed <ahmedelsaeedmassad@gmail.com> |
| URL: | https://github.com/asmpro7/metagroup/ |
| Description: | Performs meaningful subgrouping in a meta-analysis. This is a two-step process; first, use the iterative grouping functions (e.g., mgbin(), mgcont() ) to partition studies into statistically homogeneous clusters based on their effect size data. Second, use the meaning() function to analyze these new subgroups and understand their composition based on study-level characteristics (e.g., country, setting). This approach helps to uncover hidden structures in meta-analytic data and provide a deeper interpretation of heterogeneity. |
| License: | GPL (≥ 3) |
| Encoding: | UTF-8 |
| Imports: | dplyr, ggplot2, magrittr, meta, rlang |
| RoxygenNote: | 7.3.2 |
| NeedsCompilation: | no |
| Packaged: | 2025-09-05 13:15:53 UTC; ASM PRO |
| Repository: | CRAN |
| Date/Publication: | 2025-09-10 08:00:13 UTC |
metagroup: Meaningful Grouping of Studies in Meta-Analysis
Description
The 'metagroup' package provides a suite of tools to perform meaningful subgrouping in a meta-analysis. This is a two-step process: first, use the iterative grouping functions (e.g., 'mgbin()', 'mgcont()') to partition studies into statistically homogeneous clusters. Second, use the 'meaning()' function to analyze these new subgroups and understand their composition based on study-level characteristics (e.g., country, setting).
Author(s)
Ahmed Abdelmageed ahmedelsaeedmassad@gmail.com
See Also
Useful functions:
Pipe operator
Description
See magrittr::%>% for details.
Usage
lhs %>% rhs
Arguments
lhs |
A value or the magrittr placeholder. |
rhs |
A function call using the magrittr semantics. |
Value
The result of calling 'rhs(lhs)'.
Print and Plot Methods for 'grouped' Objects
Description
S3 methods for objects of class 'grouped' returned by the 'meaning()' function.
'print.grouped' provides a concise summary view of the results.
'plot.grouped' creates a faceted bar chart to visualize the composition of each subgroup.
Usage
## S3 method for class 'grouped'
print(x, ...)
## S3 method for class 'grouped'
plot(x, ...)
Arguments
x |
An object of class 'grouped'. |
... |
Additional arguments (currently unused). |
Value
The 'print' method is called for its side-effect of printing a summary table to the console.
The 'plot' method is called for its side-effect of generating a faceted bar chart in the current graphics device.
Both methods invisibly return the original object 'x'.
Author(s)
Ahmed Abdelmageed ahmedelsaeedmassad@gmail.com
See Also
Explore Composition of Homogeneous Study Subgroups
Description
After grouping studies using a meta-grouping function (e.g., 'mgbin'), this function analyzes the composition of each subgroup based on specified categorical variables. It helps to give "meaning" to the statistically derived groups.
Usage
meaning(data, subgroup = "subgroup", variables, min = 3)
Arguments
data |
A data frame or the list object returned by a meta-grouping function. If a list, the function will automatically use the 'data' element. |
subgroup |
A string specifying the name of the subgroup column. Defaults to '"subgroup"'. |
variables |
A character vector of column names (categorical variables) to analyze. |
min |
An integer specifying the minimum number of studies a group must have to be included in the final summarized output. Defaults to 3. |
Details
For each subgroup and each specified variable, the function calculates the percentage distribution of its categories. It identifies the most frequent (dominant) category and performs a chi-squared test to assess if the distribution of categories within that group is significantly different from random.
Value
An S3 object of class 'grouped', which is a list containing three data frames:
'detailed': Contains the percentage distribution for all categories of all variables within every group.
'up.hand': A subset of 'detailed', showing only the dominant category for each variable in each group.
'final': A summarized version of 'up.hand', showing the single most dominant characteristic for each group that meets the 'min' size threshold. This is the default print output.
Author(s)
Ahmed Abdelmageed ahmedelsaeedmassad@gmail.com
See Also
Examples
# Create a sample dataset of studies
study_data <- data.frame(
author = letters[1:10],
country = sample(c("USA", "China", "Egypt"), 10, replace = TRUE),
setting = sample(c("Hospital", "Community"), 10, replace = TRUE),
subgroup = c("group 1", "group 2", "group 1", "group 1", "group 2",
"group 2", "group 1", "group 3", "group 2", "group 2")
)
# Analyze the composition of the subgroups
meaning_result <- meaning(study_data, variables = c("country", "setting"))
# The default print shows the 'final' summary table
print(meaning_result)
# Access detailed results
# meaning_result$detailed
Group Studies with Binary Outcome Data by Homogeneity
Description
This function iteratively assigns studies to subgroups based on a homogeneity test. The goal is to create statistically homogeneous groups of studies before performing a final meta-analysis with binary outcome data.
Usage
mgbin(data, event.e, n.e, event.c, n.c, studlab, ...)
Arguments
data |
A data frame containing the meta-analysis data. |
event.e |
A vector of event counts in the experimental group. |
n.e |
A vector of sample sizes in the experimental group. |
event.c |
A vector of event counts in the control group. |
n.c |
A vector of sample sizes in the control group. |
studlab |
A vector of study labels. |
... |
Additional arguments passed on to 'meta::metabin'. |
Details
The algorithm starts with a single study in "group 1". It then processes each subsequent study, attempting to place it in an existing group. A study is added to a group only if its inclusion does not result in significant within-group heterogeneity. If no suitable group is found, a new one is created.
Value
A list containing the final data with subgroup assignments ('data'), the final 'metabin' model ('model'), and the number of attempts ('attempts').
Author(s)
Ahmed Abdelmageed ahmedelsaeedmassad@gmail.com
See Also
Group Studies with Continuous Outcome Data by Homogeneity
Description
This function iteratively assigns studies to subgroups based on a homogeneity test. The goal is to create statistically homogeneous groups of studies before performing a final meta-analysis with continuous outcome data.
Usage
mgcont(data, n.e, mean.e, sd.e, n.c, mean.c, sd.c, studlab, ...)
Arguments
data |
A data frame containing the meta-analysis data. |
n.e |
A vector of sample sizes in the experimental group. |
mean.e |
A vector of means in the experimental group. |
sd.e |
A vector of standard deviations in the experimental group. |
n.c |
A vector of sample sizes in the control group. |
mean.c |
A vector of means in the control group. |
sd.c |
A vector of standard deviations in the control group. |
studlab |
A vector of study labels. |
... |
Additional arguments passed on to 'meta::metacont'. |
Details
The algorithm starts with a single study in "group 1". It then processes each subsequent study, attempting to place it in an existing group. A study is added to a group only if its inclusion does not result in significant within-group heterogeneity. If no suitable group is found, a new one is created.
Value
A list containing the final data with subgroup assignments ('data'), the final 'metacont' model ('model'), and the number of attempts ('attempts').
Author(s)
Ahmed Abdelmageed ahmedelsaeedmassad@gmail.com
See Also
Group Studies with Correlation Data by Homogeneity
Description
This function iteratively assigns studies to subgroups based on a homogeneity test. The goal is to create statistically homogeneous groups of studies before performing a final meta-analysis with correlation data.
Usage
mgcor(data, cor, n, studlab, ...)
Arguments
data |
A data frame containing the meta-analysis data. |
cor |
A vector of correlation coefficients. |
n |
A vector of corresponding sample sizes. |
studlab |
A vector of study labels. |
... |
Additional arguments passed on to 'meta::metacor'. |
Details
The algorithm starts with a single study in "group 1". It then processes each subsequent study, attempting to place it in an existing group. A study is added to a group only if its inclusion does not result in significant within-group heterogeneity. If no suitable group is found, a new one is created.
Value
A list containing the final data with subgroup assignments ('data'), the final 'metacor' model ('model'), and the number of attempts ('attempts').
Author(s)
Ahmed Abdelmageed ahmedelsaeedmassad@gmail.com
See Also
Group Studies Using the Generic Inverse Variance Method
Description
This function iteratively assigns studies to subgroups based on a homogeneity test. The goal is to create statistically homogeneous groups of studies before performing a final generic inverse variance meta-analysis.
Usage
mggen(data, TE, seTE, studlab, ...)
Arguments
data |
A data frame containing the meta-analysis data. |
TE |
A vector of treatment effects. |
seTE |
A vector of the standard errors of the treatment effects. |
studlab |
A vector of study labels. |
... |
Additional arguments passed on to 'meta::metagen'. |
Details
The algorithm starts with a single study in "group 1". It then processes each subsequent study, attempting to place it in an existing group. A study is added to a group only if its inclusion does not result in significant within-group heterogeneity. If no suitable group is found, a new one is created.
Value
A list containing the final data with subgroup assignments ('data'), the final 'metagen' model ('model'), and the number of attempts ('attempts').
Author(s)
Ahmed Abdelmageed ahmedelsaeedmassad@gmail.com
See Also
Group Studies with Incidence Rate Data by Homogeneity
Description
This function iteratively assigns studies to subgroups based on a homogeneity test. The goal is to create statistically homogeneous groups of studies before performing a final meta-analysis of incidence rate data.
Usage
mginc(data, event.e, time.e, event.c, time.c, studlab, ...)
Arguments
data |
A data frame containing the meta-analysis data. |
event.e |
A vector of event counts in the experimental group. |
time.e |
A vector of person-time at risk in the experimental group. |
event.c |
A vector of event counts in the control group. |
time.c |
A vector of person-time at risk in the control group. |
studlab |
A vector of study labels. |
... |
Additional arguments passed on to 'meta::metainc'. |
Details
The algorithm starts with a single study in "group 1". It then processes each subsequent study, attempting to place it in an existing group. A study is added to a group only if its inclusion does not result in significant within-group heterogeneity. If no suitable group is found, a new one is created.
Value
A list containing the final data with subgroup assignments ('data'), the final 'metainc' model ('model'), and the number of attempts ('attempts').
Author(s)
Ahmed Abdelmageed ahmedelsaeedmassad@gmail.com
See Also
Group Studies with Single Means by Homogeneity
Description
This function iteratively assigns studies to subgroups based on a homogeneity test. The goal is to create statistically homogeneous groups of studies before performing a final meta-analysis of single means.
Usage
mgmean(data, mean, sd, n, studlab, ...)
Arguments
data |
A data frame containing the meta-analysis data. |
mean |
A vector of means. |
sd |
A vector of standard deviations. |
n |
A vector of sample sizes. |
studlab |
A vector of study labels. |
... |
Additional arguments passed on to 'meta::metamean'. |
Details
The algorithm starts with a single study in "group 1". It then processes each subsequent study, attempting to place it in an existing group. A study is added to a group only if its inclusion does not result in significant within-group heterogeneity. If no suitable group is found, a new one is created.
Value
A list containing the final data with subgroup assignments ('data'), the final 'metamean' model ('model'), and the number of attempts ('attempts').
Author(s)
Ahmed Abdelmageed ahmedelsaeedmassad@gmail.com
See Also
Group Studies with Single Proportions by Homogeneity
Description
This function iteratively assigns studies to subgroups based on a homogeneity test. The goal is to create statistically homogeneous groups of studies before performing a final meta-analysis of single proportions.
Usage
mgprop(data, event, n, studlab, ...)
Arguments
data |
A data frame containing the meta-analysis data. |
event |
A vector of event counts. |
n |
A vector of corresponding sample sizes. |
studlab |
A vector of study labels. |
... |
Additional arguments passed on to 'meta::metaprop'. |
Details
The algorithm starts with a single study in "group 1". It then processes each subsequent study, attempting to place it in an existing group. A study is added to a group only if its inclusion does not result in significant within-group heterogeneity. If no suitable group is found, a new one is created.
Value
A list containing the final data with subgroup assignments ('data'), the final 'metaprop' model ('model'), and the number of attempts ('attempts').
Author(s)
Ahmed Abdelmageed ahmedelsaeedmassad@gmail.com
See Also
Group Studies with Single Incidence Rates by Homogeneity
Description
This function iteratively assigns studies to subgroups based on a homogeneity test. The goal is to create statistically homogeneous groups of studies before performing a final meta-analysis of single incidence rates.
Usage
mgrate(data, event, time, studlab, ...)
Arguments
data |
A data frame containing the meta-analysis data. |
event |
A vector of event counts. |
time |
A vector of person-time at risk. |
studlab |
A vector of study labels. |
... |
Additional arguments passed on to 'meta::metarate'. |
Details
The algorithm starts with a single study in "group 1". It then processes each subsequent study, attempting to place it in an existing group. A study is added to a group only if its inclusion does not result in significant within-group heterogeneity. If no suitable group is found, a new one is created.
Value
A list containing the final data with subgroup assignments ('data'), the final 'metarate' model ('model'), and the number of attempts ('attempts').
Author(s)
Ahmed Abdelmageed ahmedelsaeedmassad@gmail.com