Skip to contents

Changelog

Source: NEWS.md

growthcleanr 3.0.0

growthcleanr 3.0.0 is a major update addressing usability, cleaning performance, customization, efficiency, and documentation. This changelog lists key changes. If you used an earlier version, start with vignette("experienced-start-here"), which explains how to update 2.2.3 code and describes all of these changes in more detail. For full documentation of current behavior, see the package vignettes (browseVignettes("growthcleanr")) and growthcleanr.org.

New: Shiny app

  • growthcleanr now has a point-and-click app, so you can clean data without writing R code. Start it with growthcleanr_app(). The app walks you through loading a data file, mapping your columns, and running growthcleanr, and lets you save the results. It runs entirely on your own computer; no data are uploaded. See the Shiny app how-to guide.

Main changes

  • Algorithms. Multiple changes to improve performance, including: an updated child algorithm with improved handling of infants and cleaning of head circumference through age 5 years; an updated adult algorithm with four permissiveness levels; a new “evil twins” step to detect runs of consecutive erroneous values; rescue of many consecutive identical values that were previously excluded as carried forward; and built-in recentering medians based on clinical data.
  • Input. cleangrowth() accepts a data frame, with the option to use alternate column names. Data can be in one of three formats: long (one measurement per row), semi-wide (one visit per row), or wide (one subject per row). Users may enter date of birth and visit date rather than agedays. An optional id column makes results fully replicable. The Shiny app also handles these formats and dates. For ready-to-run scripts, see vignette("example-scripts"); for full detail, see “Input format: long, semi-wide, and wide” and “Date parsing” in vignette("wrapper").
  • Output. cleangrowth() returns your data with results added. New outputs include redesigned exclusion codes and bin_exclude, unit_error_range, and mean_ht columns.
  • Efficiency. Checkpointing, on by default for datasets with more than 5,000 subjects, reduces memory use and keeps run time roughly linear in the number of subjects. Parallel processing is simpler to use (parallel = TRUE). gc_partial_run() re-cleans only subjects whose data changed.
  • Trial runs. gc_runtrial() runs growthcleanr on a sample of your data and reports how your columns, sex coding, and parameters were read, so you can check your setup before a full run.
  • Documentation. How-to guides for the Shiny app and the R interface, a researcher overview, a fact sheet, ready-to-run scripts (vignette("example-scripts")), and expanded technical documentation.

Breaking changes

Code written for 2.2.3 will need to be updated. We apologize for any inconvenience this causes to experienced users but hope these changes make growthcleanr easier for most users. Details and replacement code are in “Updating code written for growthcleanr 2.2.3” in vignette("experienced-start-here").

  • Results differ from 2.2.3, and no setting reproduces 2.2.3 exactly. To reproduce an analysis done with 2.2.3, keep using 2.2.3.
  • cleangrowth() takes a data frame as its first argument and returns a data.table, not a vector of exclusion codes. The content of the old vector is now in the exclude column. Old calls with unnamed arguments stop with a message, but code that names its arguments and treats the result as a vector can run without an error and give wrong results.
  • Exclusion codes were renamed to be more meaningful. Filters on specific 2.2.3 codes will not match anything, without a warning. See “Mapping from prior exclusion codes” in vignette("experienced-start-here").
  • growthcleanr now excludes only a subset of consecutive identical values, controlled by the new cf_rescue argument. testacf() and acf_answers() were removed, and adjustcarryforward() now only stops with a message. See vignette("cf-rescue").
  • Several arguments were removed or renamed, and a removed argument stops with a message naming what to use instead. For example, num.batches is now cores, and weight_cap in kg is now adult_scale_max_lbs in pounds. See “Removed and renamed arguments” in vignette("experienced-start-here").
  • The sd.recenter argument still allows users to enter their own recentering table. With the default, sd.recenter = NA, growthcleanr now uses a built-in recentering table based on clinical data. The "nhanes" and "derive" options and sd_median() were removed. See vignette("recentering").
  • splitinput() was removed; checkpointing replaces it. See vignette("large-datasets").
  • The command-line driver moved (find it with gc_cli_path()) and accepts only long flags. See vignette("command-line").
  • ext_bmiz() output keeps your column names, and longwide() arguments changed. To avoid errors related to numeric sex coding, longwide() now returns sex as M or F, not numeric sex, so filters of longwide() output on sex == 1 or sex == 2 will match no rows, without a warning. ext_bmiz() now stops on numeric sex unless sex has both 1s and 2s and no other values. See vignette("utilities").
  • Docker support was dropped.

Other changes

growthcleanr 2.2.3 - 2026-05-03

CRAN release: 2026-05-09

Changed

  • Fixed bug affecting adult_cutpoint affecting datasets with narrow age ranges
  • Corrected a problem with reference values

growthcleanr 2.2.2 - 2026-02-20

Changed

  • Removed outdated statement from README re bug that has now been fixed.

growthcleanr 2.2.1 - 2026-02-18

CRAN release: 2026-02-18

Changed

  • Updated internal dplyr code to use .data$ pronoun notation for CRAN compatibility with recent dplyr versions, resolving “no visible binding for global variable” notes
  • Added explicit rlang dependency for .data pronoun support
  • Fixed recentering table that had previously erroneously caused birth measurements to be excluded in small datasets
  • Updated test expectations to reflect corrected recentering behavior
  • Resolved remaining “no visible binding” note for id variable in data.table operations
  • Updated broken CDC SAS program URL to current location

growthcleanr 2.2.0-prelim-infants - 2023-09-13

CRAN release: 2024-02-26

Added

  • Added option to cleangrowth for the preliminary infants algorithm – expands pediatric algorithm to consider 0 - 2 years old, with infants = TRUE in cleangrowth(), with all steps implemented. Note that this option is still preliminary and should not be used for research. For more information regarding the logic of the algorithm, see the vignette ‘Preliminary Infants Algorithm.’

growthcleanr 2.1.1 - 2023-03-01

CRAN release: 2023-03-03

Changed

  • Fixed missing adult measurements to be labeled “Missing” in output (#119)
  • Added tests for missingness in adult output
  • Fixed missing “-RV” codes in adult output
  • Corrected contributor names in DESCRIPTION (#120)
  • Added email for Dan Chudnov in DESCRIPTION (#95)

growthcleanr 2.1.0 - 2023-02-03

CRAN release: 2023-02-16

Added

  • Use dependabot to update GitHub workflow action versions (#94)
  • Use GitHub action to build and publish container image (#101)

Changed

  • Updated ext_bmiz() to match Dec 2022 NCHS guidelines (#98)
  • New options to keep dates, columns, unmatched rows in longwide() (#71)
  • Updated CITATION to match new CRAN requirements
  • Updated Dockerfile to build from repo, not CRAN

growthcleanr 2.0.3 - 2022-11-01

CRAN release: 2022-11-03

Added

  • CRAN release checklist now added under Developer Guidelines vignette (#99)

Changed

  • All possible levels for cleangrowth() output factor now enumerated
  • Updated maintainer to Carrie Daymont

growthcleanr 2.0.2 - 2022-09-13

CRAN release: 2022-09-13

Added

Changed

  • Several updates for CRAN deployment: improved example/test runtimes, text corrections (#82); switched examples to use donttest, added CRAN comments file, updated .Rbuildignore (#84)
  • Documentation updated with CRAN install (#86), fixed links (#85)
  • Updated GitHub check workflow (#80) and pkgdown workflow

growthcleanr 2.0.1 - 2022-08-29

CRAN release: 2022-09-01

Changed

  • Updated DESCRIPTION, including authors, URLS, title, description, and imports
  • Compressed files in inst/extdata for size requirements; added R.utils as import to support fread() for .gz files
  • Updated license year

growthcleanr 2.0.0 - 2021-06-30

Added

Changed

growthcleanr 1.2.6 - 2021-06-10

Changed

  • Corrected four duplicated age-rows in NHANES reference medians (#40)
  • Added missing non-newborn constraint in 14h.ii (thanks Lusha Cao)
  • Removed Hmisc dependency (#36)
  • Replaced clean_value result column name in docs with gcr_result for clarity (#35)

growthcleanr 1.2.5 - 2021-02-26

Added

  • Added inst/extdata/nhanes-reference-medians.csv, reference medians for recentering derived from NHANES (described in README)

Changed

growthcleanr 1.2.4 - 2021-01-14

Changed

  • Minor update to WHO HT velocity 3SD files to correct a small number of errors (#24). Affected files were:

    • inst/extdata/who_ht_maxvel_3sd.csv
    • inst/extdata/who_ht_vel_3sd.csv

    Although these changes were very minor, it is possible that results on data cleaned after this change may vary from previous results. The prior version of these files may be obtained by visiting the tagged release version 1.2.3 at https://github.com/growth-data-tools/growthcleanr/releases/tag/1.2.3.

    The released version of growthcleanr available at that link contains the older version of both files; that older version may be used to verify reproducibility.

    Alternatively, a more recent version of growthcleanr may be used with only the affected files replaced with their older versions available at the 1.2.3 tag link above. This must be done manually.

growthcleanr 1.2.3 - 2021-01-07

Added

  • New exclusion handling option on experimental carry forward adjustment

Changed

  • Improved experimental carry forward adjustment handling of strings of CF values, output handling, and documentation; renamed “Missing” values
  • Updated DESCRIPTION, imports, documentation to address testing issue (#12)
  • Switched to R-native argparser library to support script options
  • Switched to GitHub Actions for continuous integration / testing (thanks
  • Improved Dockerfile to standardize user/path, simplify install (thanks

growthcleanr 1.2.2 - 2020-09-29

Added

  • CITATION file, now citation("growthcleanr") works as expected

Changed

  • Standardized on arrow assignment
  • Moved functions previously within other functions to top level
  • @import now preferred over library() for library loading
  • Exported more functions
  • Improved carried forward adjustment driver script, now supports line-grid (like original sweep), random, and grid-search search types, with configuration
  • Added fdir option to splitinput() to specify split file directory
  • Added package minimum versions to DESCRIPTION
  • Fixed example code to reduce build warnings
  • Improved and corrected documentation
  • Re-compressed synthetic sample data (syngrowth) to improve compression

growthcleanr 1.2.1 - 2020-08-14

Added

  • New tests in tests/testthat/test-utils.R and tests/testthat/test-cdc.R to support newly added functions

Changed

  • Improved error handling in longwide(); fixed missing import in DESCRIPTION

growthcleanr 1.2 - 2020-07-24

Added

Changed

  • Reorganized code from R/growth.R into separate files for clarity and easier maintenance (all utility functions not directly used by cleangrowth() are now in R/utils.R)
  • Updated README with details and examples for added functions

growthcleanr 1.1 - 2020-02-07

Added

  • New options to add flexibility:
    • error.load.mincount and error.load.threshold
    • lt3.exclude.mode with default (same as before) and flag.both mode for handling unmatched pairs
    • sdmedian.filename and sdrecentered.filename
  • New splitinput() function
  • New example synthetic data set syngrowth loads automatically.

Changed

  • Several updates to improve performance, including eliminating use of data.table in ewma function.
  • Updated README with link to paper, detailed introduction, more installation details, examples, notes on handling large datasets, lists of parameters and exclusions.

growthcleanr 1.0.0 - 2018-09-11

Added

  • Initial version posted to GitHub.