Skip to contents

Official Git repository of tracenma R package

Website

The website of tracenma currently includes a structured reference list to facilitate access to the documentation of the functions available in the package.

Description

The tracenma R package contains a database with extracted aggregate study-level characteristics (that may act as effect modifiers) from 217 systematic reviews with network meta-analysis published from 2004 to 2015. tracenma also contains functions to access the catalogue with the available systematic reviews and the datasets with the extracted characteristics, comprising the database.

The package is built upon the R package nmadb (version 1.2.0) to define the eligible connected networks to extract the available characteristics from the corresponding systematic reviews. Currently, tracenma includes a subset of the systematic reviews available in nmadb.

👉 The tracenma R package should only be used to develop and appraise the methodology to assess the transitivity assumption quantitatively.

👎 The tracenma R package should not be used, for instance, to map the characteristics reported in specific healthcare fields of the database.

Getting started

Run the following code to install and load the package from CRAN:

install.packages("tracenma")
library(tracenma)

or run the following code to install and load the development version of the package:

install.packages("devtools")
devtools::install_github("LoukiaSpin/tracenma")
library(tracenma)

Example

Access the catalogue of the database

To access the complete catalogue with all 217 systematic reviews with their characteristics, type index.

head(index[, 1:5])
#>   nmadb.ID PMID      First.Author Year    Journal.Name 
#> 1   501330 15147585         Mason 2004   BMC Fam Pract
#> 2   479574 15820294 Biondi-Zoccai 2005   Int J Cardiol
#> 5   501268 15968013  Gafter-Gvili 2005  Ann Intern Med
#> 4   481193 16354303         Moore 2005        BMC Urol
#> 3   479688 16442888          Zhou 2005      Am Heart J
#> 6   501211 16239897        Berner 2006 Int J Impot Res

Access the index of datasets published during a specific year

Use the function get.dataset.index to get one or more systematic reviews and their characteristics. For instance, let us access the available systematic reviews publish in 2007 using the default arguments:

get.dataset.index(pmid = NULL, year = 2007)
#>    nmadb.ID     PMID  First.Author Year          Journal.Name   Outcome.Type
#> 11   480655 16951908   Vestergaard 2007        Osteoporos Int      Objective
#> 14   501355 17478472         Nixon 2007          Rheumatology     Subjective
#> 13   501336 17651658        McLeod 2007 Health Technol Assess     Subjective
#> 15   501395 17903393 Soares-Weiser 2007 Health Technol Assess Semi-objective
#> 12   501309 17932160           Lam 2007                   BMJ      Objective
                                                                                                                                                
#>    Intervention.Comparison.Type Includes.ToC.where                      Source.ToC
#> 11   pharmacological vs placebo          Main text                         Table 1
#> 14   pharmacological vs placebo          Main text                         Table 1
#> 13   pharmacological vs placebo          Main text                    Table 2 to 4
#> 15   pharmacological vs placebo          Both      Tables 1 & 2, Appendix Tables 5
#> 12   non-pharmacological vs any          Main text                         Table 1

You can access the article titles by adding the argument show.title = TRUE. Then the column Title will appear after Journal.Name. Add the argument show.comment = TRUE to activate the extraction-related comments accompanying the articles. The column Comment will appear after Source.ToC. The default for both arguments is FALSE (do not report).

Access the index of specific datasets

Now, let us access the systematic reviews with PMID numbers 16951908 and 17932160 (again, using the default arguments):

get.dataset.index(pmid = c(16951908, 17932160))
#>    nmadb.ID     PMID  First.Author Year    Journal.Name  Outcome.Type
#> 11   480655 16951908   Vestergaard 2007  Osteoporos Int     Objective
#> 12   501309 17932160           Lam 2007             BMJ     Objective
                                                                                                                                                    
#>    Intervention.Comparison.Type Includes.ToC.where Source.ToC
#> 11   pharmacological vs placebo          Main text    Table 1
#> 12   non-pharmacological vs any          Main text    Table 1

Access a specific dataset

Use the function get.dataset to get the dataset with the extracted characteristics of a specific systematic review. For instance, let us access the dataset of the systematic review with PMID number 16951908 using the default arguments:

get.dataset(pmid = 16951908)
#> # A tibble: 25 × 12
#>    `trial (reference)`   treat1 treat2    arm1    arm2    sex sample.size h.rPTH   calcium vitamin.D duration quality
#>    <chr>                  <dbl>  <dbl>   <chr>   <chr>  <chr>       <dbl>  <chr>     <dbl>     <dbl>    <dbl>   <dbl>
#>  1 Finkelstein 1998 (26)      1      2 control  PTH 40 female          43   1-34        NA        NA       12       2
#>  2 Lane 1998 (42)             1      3 control  PTH 25 female          51   1-34      1500       800       12       3
#>  3 Kurland 2000 (33)          1      4 control  PTH 32   male          23   1-34      1500       400       18       3
#>  4 Cosman 2001 (24)           1      3 control  PTH 25 female          52   1-34      1500       800       36       3
#>  5 Neer 2001 (5)              1      2 control  PTH 40 female         882   1-34      1000       400       21       3
#>  6 Neer 2001 (5)              1      5 control  PTH 20 female         892   1-34      1000       400       21       3
#>  7 Neer 2001 (5)              2      5  PTH 40  PTH 20 female         878   1-34      1000       400       21       3
#>  8 Body 2002 (43)             1      2 control  PTH 40 female         146   1-34      1000       400       12       4
#>  9 Finkelstein 2003 (28)      1      2 control  PTH 40   male          48   1-34      1000       400       30       3
#> 10 Finkelstein 2003 (28)      1      6 control PTH+ALN   male          53   1-34      1000       400       30       3
#> # ℹ 15 more rows
#> # ℹ Use `print(n = ...)` to see more rows

Adding the argument show.index = TRUE returns an additional data-frame (following the dataset, which is skipped in the example below) that presents the abbreviation (left column) and full name (right column) of each characteristic:

get.dataset(pmid = 16951908, show.index = TRUE)
#> $Characteristics_index
#> # A tibble: 7 × 2
#>   Abbreviation Full.name                              
#>   <chr>        <chr>                                  
#> 1 sex          sex                                    
#> 2 sample.size  sample size                            
#> 3 h.rPTH       human (recombinant) parathyroid hormone
#> 4 calcium      calcium in mg/day                      
#> 5 vitamin.D    vitamin D in IU/day                    
#> 6 duration     trial duration in months               
#> 7 quality      study quality  

Lastly, adding the argument show.type = TRUE enhances the additional data-frame, $Characteristics_index, with another two columns appearing on the right; the Characteristic.type (Clinical, Demographic, Methodological) and Characteristic.subtype (Participant, Intervention, Outcome, Age, Sex, Ethnicity, Study design, Study setting, Risk of bias, Withdrawals) for each characteristic:

get.dataset(pmid = 16951908, show.index = TRUE, show.type = TRUE)
$Characteristics_index
#> # A tibble: 7 × 4
#>   Abbreviation Full.name                               Characteristic.type Characteristic.subtype
#>   <chr>        <chr>                                   <chr>               <chr>                 
#> 1 sex          sex                                     Demographic         Sex                   
#> 2 sample.size  sample size                             Methodological      Study design          
#> 3 h.rPTH       human (recombinant) parathyroid hormone Clinical            Intervention          
#> 4 calcium      calcium in mg/day                       Clinical            Intervention          
#> 5 vitamin.D    vitamin D in IU/day                     Clinical            Intervention          
#> 6 duration     trial duration in months                Methodological      Study design          
#> 7 quality      study quality                           Methodological      Risk of bias          

The default for both arguments is FALSE (do not report the data-frame $Characteristics_index).

That’s it for the moment! 😎 ☕

Acknowledgements

I would like to thank Juan Jose Yepes-Nuñez and Andrés Mauricio García-Sierra for their valuable help during data curation of the extracted datasets for possible extraction errors.

Funding source

The development of the tracenma R package is funded by the German Research Foundation (Deutsche Forschungsgemeinschaft) (grant no. SP 1664/2-1)