Tools for reading the GSED validation data • gsedread

The goal of gsedread is to read the Phase 1 and Phase 2 validation data of the project Global Scales for Early Development (GSED).

Installation

Install the gsedread package from GitHub as follows:

install.packages("remotes")
remotes::install_github("d-score/gsedread")

How to use

You need access to the proper SharePoint site and sync the data to a local OneDrive. In the file .Renviron in your home directory add a line specifying the location of your synced OneDrive, e.g.,

ONEDRIVE_GSED='/Users/username/Library/CloudStorage/OneDrive-Sharedlibraries-...'

After saving .Renviron, (re)start R, and manually check whether you are able to read the OneDrive directory. The following R expression should return the contents of the OneDrive directory:

head(dir(Sys.getenv("ONEDRIVE_GSED")))
#> [1] "-DESKTOP-GU6P9PF.RData"                       
#> [2] "-DESKTOP-GU6P9PF.Rhistory"                    
#> [3] "Bangladesh Validation"                        
#> [4] "Baseline Analysis - OLD - NOV 2021"           
#> [5] "Data Cleaning Script MK1 - Run before merge.R"
#> [6] "DATA DICTIONARY"

Then set the proper paths to and within the OneDrive directory where the data are located. For example, for Phase 1 data, the paths are:

onedrive <- Sys.getenv("ONEDRIVE_GSED")
path <- file.path("GSED Phase 1 Final Analysis",
                  "GSED Final Collated Phase 1 Data Files 18_05_22")

The high-level function read_gsed_fixed() reads and combines SF, LF and BSID data files. For the Phase 1 data, the function is called as follows:

library(gsedread)
phase1 <- read_gsed_fixed(onedrive, path, phase = 1)

The read_gsed_fixed() functions returns a list with two elements: visits and responses.

The visits element is a data frame with the visit information, with the following structure:

str(phase1$visits)
#> tibble [12,888 × 27] (S3: tbl_df/tbl/data.frame)
#>  $ file                 : chr [1:12888] "ban_sf_2021_11_03" "ban_sf_2021_11_03" "ban_sf_2021_11_03" "ban_sf_2021_11_03" ...
#>  $ subjid               : chr [1:12888] "11-GSED-0001" "11-GSED-0001" "11-GSED-0001" "11-GSED-0002" ...
#>  $ parent_id            : chr [1:12888] "627662" "627662" "627662" "618677" ...
#>  $ worker_code          : num [1:12888] 224 224 223 227 227 223 223 225 228 226 ...
#>  $ vist_type            : int [1:12888] 1 6 8 1 6 8 1 1 1 1 ...
#>  $ location             : num [1:12888] 1 1 1 1 1 1 1 1 1 1 ...
#>  $ agedays              : int [1:12888] 421 429 631 581 589 791 85 245 516 176 ...
#>  $ ah01                 : num [1:12888] NA 1 1 NA 1 1 NA NA NA NA ...
#>  $ ah02                 : num [1:12888] NA 0 0 NA 0 0 NA NA NA NA ...
#>  $ ah03                 : num [1:12888] NA 0 0 NA 0 0 NA NA NA NA ...
#>  $ ah04                 : num [1:12888] NA 0 0 NA 0 0 NA NA NA NA ...
#>  $ ah05                 : num [1:12888] NA 0 0 NA 0 0 NA NA NA NA ...
#>  $ ah06                 : num [1:12888] NA 0 0 NA 0 0 NA NA NA NA ...
#>  $ ah07                 : num [1:12888] NA 0 0 NA 0 0 NA NA NA NA ...
#>  $ ah08                 : num [1:12888] NA 0 0 NA 0 0 NA NA NA NA ...
#>  $ m01                  : num [1:12888] NA NA 3 NA NA 2 NA NA NA NA ...
#>  $ m02                  : num [1:12888] NA NA 0 NA NA 0 NA NA NA NA ...
#>  $ m03                  : chr [1:12888] NA NA NA NA ...
#>  $ date                 : Date[1:12888], format: "2021-02-10" "2021-02-18" ...
#>  $ caregiver            : num [1:12888] 11 11 11 11 11 11 11 11 11 11 ...
#>  $ adm                  : chr [1:12888] "fixed" "fixed" "fixed" "fixed" ...
#>  $ ins                  : chr [1:12888] "sf" "sf" "sf" "sf" ...
#>  $ d01                  : int [1:12888] NA NA NA NA NA NA NA NA NA NA ...
#>  $ d01_spcy             : chr [1:12888] NA NA NA NA ...
#>  $ d02                  : int [1:12888] NA NA NA NA NA NA NA NA NA NA ...
#>  $ d02_spcfy            : chr [1:12888] NA NA NA NA ...
#>  $ agedays_adj_premature: int [1:12888] NA NA NA NA NA NA NA NA NA NA ...

There are 12888 visits in the data, with 4374 unique GSED IDs. A visit is defined by a combination of subjid, agedays, vist_type and ins.

The responses element is a data frame with the responses, with the following structure:

str(phase1$responses)
#> tibble [566,504 × 5] (S3: tbl_df/tbl/data.frame)
#>  $ subjid   : chr [1:566504] "11-GSED-0001" "11-GSED-0001" "11-GSED-0001" "11-GSED-0001" ...
#>  $ agedays  : int [1:566504] 421 421 421 421 421 421 421 421 421 421 ...
#>  $ vist_type: int [1:566504] 1 1 1 1 1 1 1 1 1 1 ...
#>  $ item     : chr [1:566504] "gpamoc058" "gpalgc059" "gpamoc060" "gpamoc061" ...
#>  $ response : int [1:566504] 1 1 1 1 1 1 1 1 1 1 ...

There are 566504 responses in the data, with 4374 unique GSED IDs. A response is defined by a combination of subjid, agedays, vist_type and item. The column named response is the 0/1 score on the item.

Lower-level functions

You can assess lower-level functions for reading data files, such as the Short Form (SF), by the read_sf() function. For example, to read the Phase 1 SF data files:

sf <- read_sf(onedrive, path)

The resulting dataset is in a form much closer to the raw data. Normally, you would not need to use this function, as the read_gsed_fixed() function reads and processes the data for you, but it can help to diagnose issues with the data. Similar functions exist for the Long Form (LF) and BSID data files, read_lf() and read_bsid(), respectively.

Operations

The package reads and processes GSED data. It does not store data. The read_sf(), read_lf() and read_bsid() functions take the following actions:

Construct the paths to the files OneDrive sync file
Read all specified datasets in a list
Internally specofy the desired format for each column
Specify the available date and data-time formats per file
Recode empty, NA, -8888, -8,888.00 and -9999 values as NA
Repair problems with mixed data-time formats in the adaptive Pakistan data
Stack the datasets to one tibble and adds columns file and adm
Remove records without a GSED_ID

In addition, the read_gsed_fixed() function performs the following actions:

Account for the different item orders between Phase 1 and Phase 2 for SF and LF by means of the phase argument
Add the ins field to visits, providing a short hand for instrument ("sf", "lf", "bsid")
Transform the data for consistency, efficiency and clarity
Remove duplicates and sorts the data
Rename all item names to conform to the 9-position GSED convention
Split the data into visits and responses data frames, where visits contains the visit information and responses contains the item responses.
Remove gpamoc008 (Clench fist)
Remove responses for gtolgd002, gtolgd003, gtolgd004, gtolgd006, gtolgd007 and gtolgd008 after the age of six months
Make sure that all responses are coded as 0 or 1.

Item renaming with rename_variables() relies on the item translation table at https://github.com/D-score/gsedread/blob/main/inst/extdata/itemnames_translate.txt.

Additional cleaning and data repair

The read_gsed_fixed() reads the data per phase. After combining the data from the different phases, the functions repair_visits() and repair_responses should be run to ensure that the data are in a consistent format.

Data governance

The Global Scales of Early Development (GSED) study is an international project coordinated by the World Health Organization (WHO). The data are stored in a secure SharePoint site and are shared within the GSED team for research purposes only. Team members are not allowed to share the data with third parties. See https://www.who.int/publications/i/item/WHO-MSD-GSED-package-v1.0-2023.1 for information on the GSED study.

Acknowledgement

This study was supported by the Bill & Melinda Gates Foundation. The contents are the sole responsibility of the authors and may not necessarily represent the official views of the Bill & Melinda Gates Foundation or other agencies that may have supported the primary data studies used in the present study.