README

R package for cleaning pediatric and adult weight, height, and head circumference data from electronic health records.

Version 3 is a major update. If you are new to growthcleanr, start with vignette("new-start-here"); if you have used growthcleanr before, start with vignette("experienced-start-here"). The full documentation is available at growthcleanr.org.

This package implements a modified version of the Daymont et al. algorithm, as specified in Supplemental File 3 within the Supplementary Material published with that paper. It includes significant updates that have yet to be published in a peer-reviewed journal, but are described in detail in the package documentation.

This package also includes a function for calculating percentiles and Z-scores of pediatric growth observations using the methods in CDC software and the cdcanthro R package.

Installation

From CRAN

install.packages("growthcleanr")

GitHub version

To install the current version from GitHub with remotes, building the vignettes so the in-package documentation is available locally:

install.packages("remotes")
remotes::install_github("growth-data-tools/growthcleanr",
                        build_vignettes = TRUE, dependencies = TRUE)

build_vignettes = TRUE requires the knitr and rmarkdown packages, which dependencies = TRUE installs for you. (pak::pak("growth-data-tools/growthcleanr") also works and is faster, but does not build the vignettes.)

Summary

The growthcleanr package processes weight, height, and head circumference data to identify implausible measurements and other values that should be excluded from analysis. It uses a variety of techniques, many of which rely on a comparison of a value to an expected value derived from the patient’s other values.

Results from growthcleanr include a flag for each measurement indicating a recommendation to include it or exclude it from analysis. No values are removed. These flags can be used as-is to identify an analytic dataset, but researchers can also examine detailed results to customize use of growthcleanr. The exclusion flags also facilitate reporting the reason for excluding a measurement.

To start running growthcleanr, an R installation is required, as is a growth measurement dataset prepared for use in growthcleanr. growthcleanr can be run from R, from the command line, or through a point-and-click Shiny app (growthcleanr_app()) that doesn’t require writing R code.

Documentation

The full documentation is available at growthcleanr.org: guides for getting started, configuration and output, and technical references for the child and adult algorithms. The vignettes also ship prebuilt with the package; view them locally with browseVignettes("growthcleanr").

Changes

Contributors and funding

The people and organizations who have contributed to and supported growthcleanr are listed on the Contributors page.