Skip to contents

Changelog

Source: NEWS.md

summata 0.13.0 (2026-10-18)

  • Speed up desctable(), which now computes rank-test p-values from a single sort and derives the grouping structure once per table
  • Pass the arguments in ... of desctable() (e.g., var.equal = TRUE) to the tests that accept them, with a warning for arguments no test accepts; exclude missing and empty categories from categorical tests, falling back to a Monte Carlo p-value when Fisher’s exact test exceeds its workspace
  • Fix categorical totals, median survival intervals at nondefault confidence levels, and the formatting of values of 1,000 or more in desctable(), and show the second level of binary variables in condensed tables
  • Compute the Gehan-Breslow, Tarone-Ware, and Peto-Peto tests that survtable(test_type = ) names, honor p_digits, list groups in order of first appearance as desctable() does, and match survival time quantiles to their groups when a factor level is empty
  • Show logical predictors by level, with FALSE as the reference, in regression tables and forest plots
  • Accept non-syntactic variable names (e.g., "tumor stage"), with or without backticks, in fit(), uniscreen(), multifit(), fullfit(), and survtable(), and read factor levels and variable names that contain a colon correctly
  • Fix the matching, counts, and labels of interaction terms whose variables have non-syntactic, overlapping, or logical names
  • Order the variables of the tables of models passed to fit(model = ) and m2dt() as the model formula does, and describe mixed models as univariable or multivariable from their fixed-effect terms
  • Count the events of Cox models from the event argument of Surv(), coded as Surv() codes it; counts were wrong for status coded 1 and 2, counting-process responses, clogit() models, and events written as expressions
  • Omit aliased coefficients from forest plots and Cox regression tables, show a factor whose first level is empty against its effective reference level, and drop empty factor levels before fitting Cox models
  • Compute forest plot intervals as the regression tables do, following summata.conf_method, at the requested conf_level in uniforest(), and with the t quantile for linear models
  • Fix forest plot interaction rows under indent_groups and condense_table, coxforest() intervals and interactions for coxme models, and the fixed-effect terms of mixed models (e.g., log(age)) in glmforest() and lmforest(); draw forest plots when knitr is not installed
  • Report only the predictor’s coefficients in multifit(), with observations rather than events as n for Cox models and the given order of outcomes and levels under columns = "both"; fit Cox models to Surv() outcomes, as fit() does; draw the table’s estimates in multiforest()
  • Name the predictor or outcome of a failed fit in uniscreen() and multifit(), and use one core when the number of cores cannot be determined
  • Compare linear mixed models in compfit() on maximum likelihood fits, include the Brier score in the composite score of logistic models, score the Cox global p-value independently of the number format, accept candidates with aliased coefficients, and skip profile intervals that are not displayed; use model-level N and Events in compfit() and the fit() print method
  • Bold significant p-values in exported tables under every number format, and keep the p-values of factor levels in indented and condensed exports
  • Compute the c-statistic without pROC and name GLM families without stringr; both are removed from Suggests
  • Documentation and testing suite revisions

summata 0.12.0 (2026-07-31)

CRAN release: 2026-08-21

  • Fix n/Events counts to report complete cases across all functions, so group sizes sum to the model n; empty factor levels now report zero, and multifit() adjusted columns match fullfit() (STROBE item 12)
  • Report the analyzed sample and events in print() methods and forest plot footers, with counts available via the analysis_counts attribute
  • Import Surv(), strata(), cluster(), and coxph() from survival to fix clogit and coxme failures when the package is not attached by the user
  • Fix coefficient parsing for variable names that share a prefix or contain regular expression metacharacters
  • Fix uniforest() to draw factor levels in their established order rather than alphabetically when indent_groups = TRUE
  • Fix counts of 100,000 or greater displaying in exponential notation, and centralize count formatting in a vectorized format_count()
  • Carry variable labels from the regression functions through to the forest plot functions, so labels supplied once apply to both table and plot
  • Add forestsave() for saving forest plots at their recommended dimensions, with format-appropriate graphics devices and optional font embedding; rec_dims now records its units
  • Add recdims() function for directly retrieving the figure dimensions recommended for a forest plot
  • Rename autotable() to tablesave(), pairing it with forestsave() and distinguishing it from autoforest(), which detects model class rather than file format; autotable() is retained as a deprecated alias
  • Add quiet argument to tablesave() and the table2*() functions, and return the file path invisibly from table2pdf(), table2html(), and table2tex()
  • Add table_data attribute to the forest plot functions for access to the values drawn in the figure
  • Add compfit() warning when candidate models are fitted to different numbers of observations
  • Suppress spurious diagnostics: singular-fit messages from mixed-effects screening, the deprecated lme4::findbars() call, the MuMIn::r.squaredGLMM() null model advisory, and the Hosmer-Lemeshow test where fitted values are too few to bin
  • Declare parallel, tools, and utils in Imports, and add SystemRequirements for the optional external tools
  • Documentation and testing suite revisions

summata 0.11.5 (2026-05-06)

CRAN release: 2026-05-07

  • Explicitly define %||% operator for backward compatibility
  • Move main documentation to Codefloe Pages
  • Minor documentation edits

summata 0.11.4 (2026-03-14)

CRAN release: 2026-03-20

  • Use profile likelihood CIs for GLM/negbin and exact t-distribution CIs for linear models, replacing Wald approximation
  • Add conf_method parameter ("profile" / "wald") to fit(), uniscreen(), fullfit(), multifit(), and m2dt(), with global option summata.conf_method
  • Report complete-case n/Events in fullfit() multivariable rows (STROBE item 12)
  • Cache profile likelihood CIs from fit() for reuse in forest plot functions
  • Resolve family = "Gamma" string to Gamma(link = "log") for interpretable multiplicative effects
  • Fix survival::strata() namespace in conditional logistic regression test
  • Update benchmarks feature comparison articles

summata 0.11.3 (2026-03-03)

CRAN release: 2026-03-08

  • CRAN initial release re-submission
  • Fix table2pdf() to specify output directory for files
  • Minor WORDLIST changes

summata 0.11.2 (2026-03-02)

  • CRAN initial release re-submission
  • Minimal executable examples added to all exported functions, with lengthier demonstrations wrapped in \donttest{}
  • Modify file-writing examples in documentation headers to write to tempdir()

summata 0.11.1 (2026-02-22)

  • CRAN initial release re-submission
  • Documentation LaTeX fixes
  • Broken link fixes

summata 0.11.0 (2026-02-21)

  • CRAN initial release submission
  • Upload documentation

summata 0.10.9 (2026-02-20)

  • Documentation headers fine editing and revisions
  • Minor bugfixes

summata 0.10.8 (2026-02-09)

  • Add compatibility with US/EU locales and different big/small number marks
  • Add quiet vs. verbose settings to regression functions for warnings
  • Documentation and vignette updates

summata 0.10.7 (2026-02-05)

  • Add R-hub compatibility
  • Documentation and vignette updates

summata 0.10.6 (2026-01-20)

  • Fix “n” and “Events” counts in regression summary tables for extended GLM families
  • Improve clogit example in “Regression Modeling” vignette
  • Fix errors in condense_table logic in table export and forest plot functions
  • Fix variable_padding in table export functions, ensure consistent behavior with zebra_stripes
  • Lighten forest plot colors
  • Change n/N labeling for consistency
  • Expanded documentation examples for interaction effects, mixed effects, and extended GLM families
  • Diversify use of random parameter in regression models
  • Documentation cleaning

summata 0.10.5 (2026-01-18)

  • Add count outcomes (ae_count and fu_count) to clintrial mock dataset
  • Fix color coding in forest plots
  • Fix glmforest() to correctly extract values from MASS::glm.nb()
  • Fix regression functions to respect level order in categorical functions
  • Fix bug in uniscreen() where specified p_threshold values were not generating screened outputs
  • Fixes to print outputs in regression functions
  • Multiple multifit() revisions, including improved “n” and “Events” column handling and interaction effect formatting

summata 0.10.4 (2026-01-15)

  • Revise and standardize documentation
  • Add linting fixes

summata 0.10.3 (2026-01-12)

  • Add vignettes
  • Add GitHub Actions CI/CD workflows

summata 0.10.2 (2026-01-12)

  • Add support for various GLM families
  • Add MASS dependency for MASS::glm.nb()
  • Testing suite bugfixes
  • Update documentation

summata 0.10.1 (2026-01-05)

  • Streamline intelligent condense_table string finding
  • Update documentation

summata 0.10.0 (2026-01-04)

  • Add automatic font detection to forest plot functions
  • Fix formatting for negative coefficients in forest plots
  • Fix “negative zero” formatting issues
  • Add safeguards to avoid multiple model types in multifit() calls
  • Add intelligent handling of binary categorical variables for condense_table

summata 0.9.4 (2026-01-02)

  • Add show_logs parameter to table2tex()
  • Consistent Unicode formatting for files
  • Increased R version requirement to >= 4.2.0 for better Unicode compatibility
  • Add survtable() function with utilities
  • Minor optimizations and function cleanup, with notation standardization

summata 0.9.3 (2025-12-29)

  • Package refresh, with revised naming scheme
  • README rewrite

summata 0.9.2 (2025-12-27)

  • Expand “affirmative level” string criteria for binary categorical variables
  • Add option to remove “Predictors” column in multifit() and multiforest()
  • Update documentation

summata 0.9.1 (2025-12-27)

  • Ensure user-modifiable p_digits and conf_level parameters across all regression and forest plot functions
  • Add parameter for toggling the footer in forest plot functions
  • Update uniscreen() to accept mixed-effect models
  • Update autoforest() to accept lmer and glmer objects

summata 0.9.0 (2025-12-27)

summata 0.8.3 (2025-12-24)

  • Add optimizations to descriptive statistics workflows
  • Add p_per_stat parameter to desctable()
  • Change add_reference_rows parameter to just reference_rows in regression functions

summata 0.8.2 (2025-12-23)

  • Add optimizations to regression workflows

summata 0.8.1 (2025-12-21)

  • Add cleanup to documentation headers

summata 0.8.0 (2025-12-20)

  • Add pkgdown website
  • Add testing suite
  • Add NEWS
  • Add README
  • Add vignettes
  • Add citation function

summata 0.7.14 (2025-12-20)

summata 0.7.13 (2025-12-19)

  • Rename digits_p parameter to p_digits and add to forest plot functions
  • Fix effect_label parameter in forest plot functions
  • Update/clean comments and documentation headers

summata 0.7.12 (2025-12-15)

  • Add condense_quantitative parameter to table export functions
  • Modify fit() and glmforest() to correctly display Poisson model statistics
  • Fix LaTeX/PDF export bugs

summata 0.7.11 (2025-12-13)

  • Update R version dependency (>= 4.4.0) for built-in null coalescing operator

summata 0.7.10 (2025-12-08)

  • Update MuMIn package dependency and QC statistics in lmer models
  • Ensure that lme4 and coxme models work with compfit()

summata 0.7.9 (2025-12-06)

  • Fix “n” and “Events” columns for interaction effects
  • Fix lmerMod compatibility with fit() and m2dt()
  • Fix lmer compatibility for lmforest()
  • Get interaction effects to show on forest plots

summata 0.7.8 (2025-11-24)

summata 0.7.7 (2025-11-20)

  • Add data requirement to m2dt() to allow for accurate per-group “n” and “events” columns for all models
  • Multiple R CMD check fixes

summata 0.7.6 (2025-11-12)

  • Rename var_labels parameter to just labels

summata 0.7.5 (2025-11-12)

  • Add compatibility with interaction terms

summata 0.7.4 (2025-11-03)

  • Forest plot unit conversion fixes

summata 0.7.3 (2025-11-03)

  • Add coefficient table combination function for compfit()
  • Global and imported variable fixes

summata 0.7.2 (2025-11-03)

  • Multiple R CMD check fixes

summata 0.7.1 (2025-11-03)

  • Fix global variables

summata 0.7.0 (2025-11-02)

  • Add global variables

summata 0.6.3 (2025-11-03)

  • Add suggested dependencies for vignettes and documentation

summata 0.6.2 (2025-11-03)

  • Add GitHub Actions workflow

summata 0.6.1 (2025-11-03)

  • Update internal documentation for flextable export

summata 0.6.0 (2025-11-02)

  • Add flextable table export functions with helpers
  • Documentation updates for existing table export functions

summata 0.5.12 (2025-10-31)

  • Add data.table-specific performance enhancements for core functions

summata 0.5.11 (2025-10-29)

  • Update documentation for core functions

summata 0.5.10 (2025-10-29)

  • Indent groups and condense tables logic refinements

summata 0.5.9 (2025-10-26)

  • Add zebra stripes and dark header options for all table outputs

summata 0.5.8 (2025-10-23)

  • Update forest plot functions: additive columns, zebra stripes, and units conversion

summata 0.5.7 (2025-10-18)

  • Add condensing/indenting to forest plots

summata 0.5.6 (2025-10-12)

  • Add table condensing to table export functions

summata 0.5.5 (2025-10-11)

summata 0.5.4 (2025-10-11)

  • Add clintrial sample data

summata 0.5.3 (2025-10-10)

summata 0.5.2 (2025-10-09)

  • Fix “n” and “events” columns in uscreen(), fit(), and *summata*()
  • Add hyphen space fillers for “p-value”, “Uni p”, and “Multi p” columns

summata 0.5.1 (2025-10-08)

summata 0.5.0 (2025-10-06)

summata 0.4.1 (2025-10-06)

  • Added compfit() function
  • Reorganized helper functions
  • Function name cleanup

summata 0.4.0 (2025-10-05)

  • Replacement of mmodel() function with fit()
  • Standardization of output formats

summata 0.3.1 (2025-10-02)

  • Fixes for unknown/missing rows, reporting of ranges, and non-grouped tables in desctable()

summata 0.3.0 (2025-09-30)

  • Edits to desctable() and fit()
  • Addition of “raw data” outputs
  • Expand internal documentation

summata 0.2.1 (2025-09-26)

  • Output formatting edits

summata 0.2.0 (2025-09-26)

summata 0.1.0 (2025-09-23)

  • Initial commit