gtsummary 1.6.1

New Functionality

• Added the standard error of proportions to the list of available statistics for categorical data summaries in tbl_svysummary(). (#1187)

• Added Tarone-Ware test to add_p.tbl_survfit() (#732)

• Updated add_global_p() to handle tbl_uvregression() objects where users specified the x= argument (when y= argument is more common). (#1260)

Other Updates

• Updated start-up messaging. (#1228)

• The paired.wilcox.test available in add_p.tbl_summary() and add_difference.tbl_summary() was mistakenly marked as returning a difference, but it does not. The documentation has been corrected, which results in improved messaging to the user when the test is selected in add_difference(). (#1279)

• Improved error messages for paired tests in add_p() and add_difference() when group= argument is not specified. (#1273)

• Added argument with_gtsummary_theme(msg_ignored_elements=) argument. Use this argument to message users if any theme elements will be overwritten and therefore ignored inside the with_gtsummary_theme() call. (#1266)

• Swapped gt::fmt_missing() for gt::sub_missing() as the former is now deprecated. (#1257)

• The checks for "haven_labelled" class are now only performed for the variables indicated in include= and by= in tbl_summary() and tbl_svysummary(). The checks in tbl_uvregression() and tbl_survfit.data.frame() are only applied to the variables in include=, e.g. no checking for the outcome variable(s).

• Updates to labels and default formatting functions of unweighted statistics presented in tbl_svysummary(). (#1253)

• Adding additional structural checks in tbl_merge() and inline_text() to provide better error messaging. (#1248)

• Added tbl_regression.crr() method with messaging recommending use of tidycmprsk::crr() instead. (#1237)

• The experimental support for ftExtra::colformat_md() in as_flex_table() has been removed. The function requires evaluated YAML paths and does not allow un-evaluated references like bibliography:: "`r here::here()`". (#1229)

• Update for tbl_summary(by=) that now allows for a column named "variable" to be passed. (#1234)

• Added theme element to control what missing statistic is shown in summary tables with options to display number or percent missing or non-missing or total number of observations. (#1224)

• Renamed modify_cols_merge() to modify_column_merge() to be inline with the other modify_column_*() functions.

Bug Fixes

• Fix in as_kable_extra() when output format is 'latex' where a cell that had been bold or italicized had special characters double-escaped. Added a condition not to escape special characters in these styled cells. (#1230)

• Fix in with_gtsummary_theme(). The function restored any previously set theme, but inadvertently included the temporary theme along with it.

gtsummary 1.6.0

Improvements to as_kable_extra()

• For users who used the default kableExtra print without output-specific formatting, there are no breaking changes...the only changes will be improved output styling.
• The biggest user-facing change is that the default results for LaTeX output are now greatly improved compared to previous releases, when escape=FALSE (the new default).
  • Markdown bold, italic, and underline syntax in the headers, spanning headers, caption, and footnote will be converted to escaped LaTeX code.
  • Special LaTeX characters in the body of the table will be escaped with .escape_latex() or .escape_latex2(), e.g. % will be updated to \\%, and rendered as % in the PDF document.
  • The "\n" symbol will be recognized as a line break in the table headers, spanning headers, caption, and the table body.
  • \n is removed from footnotes
  • The escape= argument is now passed to kableExtra::add_header_row() and kableExtra::footnote() as well (previously, was only passed to knitr::kable()).
• The as_kable_extra(escape=, format=) arguments have been made explicit, where previously, these arguments were passed via ....
• Breaking Change The default value of escape= is now FALSE. If users previously used as_kable_extra(escape=FALSE) and had manually escaped LaTeX/HTML characters in the body of the table, these characters will now be double escaped. To print the table without the auto-escaping that is now present, utilize the new argument as_kable_extra(addtl_fmt=FALSE)
• Breaking Change The fmt_missing= argument was added to as_tibble(), as_kable(), and as_kable_extra() in the last release. This argument is now deprecated in as_kable() and as_kable_extra(). If a user does not want missing values formatted, they can exclude these commands with the include= argument.
• Breaking Change The strip_md_bold= has been deprecated. The markdown syntax will automatically be stripped from headers, unless escape = FALSE and format = "latex". In that case, the markdown syntax will be converted to LaTeX commands.
• HTML Updates
  • The default markdown syntax in the headers and spanning headers is removed.
  • Special characters in the table body, headers, spanning headers, caption, and footnote will be escaped with .escape_html().
  • The "\n" symbol is removed from the footnotes

Improvements to as_flex_table()

• Added support for markdown syntax in {flextable} header rows by utilizing the {ftExtra} package. If this package is installed and options(gtsummary.use_ftExtra = TRUE) has been set (or the equivalent theme element), the bold/italic markdown syntax found in the headers will be styled. Otherwise, the markdown syntax is stripped from the header rows. (#1200)

New Functions

• New function as_hux_xlsx() added to export a formatted {gtsummary} table directly to Excel.

• Added a tbl_regression.tidycrr() method to summarize competing risks regression models. (#1169)

• Added tidier tidy_wald_test(), a generic function that can calculate Wald test statistics for groups of variables in a model. The tidier expects the model object is supported by both vcov() and coef() to obtain the variance-covariance matrix and the coef vector.

• Adding functions get_gtsummary_theme() and with_gtsummary_theme() for extracting the current gtsummary theme and running code with a temporarily theme.

• Added new function modify_column_indent()--a wrapper for modify_table_styling()--to make it easier to add and remove indentation in a table.

• Now exporting a function primarily used internally as a helper for converting a gtsummary table to gt (and other formats): .table_styling_expr_to_row_number().

• Theme helper function check_gtsummary_theme() has been added. The function takes a gtsummary theme list as the input, and runs various consistency checks. Useful when constructing a personalized theme. This function replaces the internal checks that a passed theme element is indeed a valid theme element.

New Functionality

• Functions bold_labels(), bold_levels(), italicize_labels(), italicize_levels() are now method functions so they can work better on tbl_cross() objects.

• Total overhaul to the way statistics are saved and reported in modify_header(), modify_spanning_header(), modify_footnote(), and modify_caption(). There is now a standardized place to these statistics to be saved in all gtsummary tables (in x$table_styling$header in columns starting with "modify_stat_"). The modify functions have been updated to access the statistics from the header data frame. An added benefit to this structure, is that the statistics are available after tables are merged and stacked. Statistics available in modify_caption() are taken from the "label" column. (#1165, #1101)

• Added add_global_p(anova_fun=) argument allowing users to pass custom functions to calculate global p-values when car::Anova() does not support the model type. (#1149)

• Functions bold_labels(), bold_levels(), italicize_labels(), and italicize_levels() now bold/italicize the first column shown in the table. Previously, the "label" column (which is most often the first shown column) was styled.

• Added theme element to pass arguments to knitr::kable() in as_kable() and as_kable_extra().

• Updated the default function in add_glance_*(glance_fun=) for MICE models. (#1137)

• Add tbl_butcher(keep=) argument to optionally keep some internal objects as needed. (#1148)

• Adding Norwegian translations (#1143)

Other Updates

• Removed use of round() in style_number(), and replaced it with a round function that does not "round-to-even". (#1140)

• Converted the Table Gallery vignette into an FAQ+Gallery (#811)

• Added error messaging if user tries to run add_p() or add_difference() twice. (#1209)

• All models CIs were labelled as a Confidence Interval. Now Bayesian models will correctly label the Credible Interval. (#1196)

• Improved error messaging the functions as_gt(), as_kable(), as_flex_table(), as_hux_table() when an object that is not class 'gtsummary' is passed. (#1188)

• Deprecated the as_huxtable(strip_md_bold=) as {huxtable} now recognizes the markdown syntax and there is no reason to remove the markdown syntax.

• Added huxtable::set_header_rows() to the as_hux_table() stack.

• Improved error messaging in modify_*() functions. (#914)

• Added style_percent() as the default formatting function for un-weighted proportions. (#1181)

• The global options previously available have now been soft deprecated. All documentation of the global options was removed in v1.3.1. (#1085)

• The as_flextable() function has been upgraded from a soft to a hard deprecation; use as_flex_table() instead.

• No longer removing the survey design columns from survey objects from the columns that will be summarized in tbl_svysummary() and those summarized in tbl_uvregression(). If users previously didn't indicate which variables to summarize with include=, then the design columns will now appear in the summary tables. (#1166)

• Added a check for functions that accept ... where nothing should be passed in the .... If a bad or misspelled argument is found, the users are informed. (#1083)

Bug Fixes

• Fix in "emmeans" methods for add_difference() to due an argument name change in the emmeans package. We now require the most recent version of the package. (#1205)

• Bug fix in inline_text.tbl_summary() where one could not pass a pattern only when the column argument was NULL. (#1193)

• Fix in as_gt() when there are no hidden columns.

• Fix in add_p.tbl_svysummary() when Wald tests were converted to flextable. The survey tidier saved the column as a matrix-column instead of a vector, which was incompatible with flextable output. (#1153)

• Fixing Lancet theme mid-point encoding issue on Linux and MacOS. (#1146)

• Fix when using summary type gtsummary selectors (e.g. all_continuous()) with the add_ci(style_fun=) argument. (#1141)

gtsummary 1.5.2

• Removed foreign reference to external functions in the top level of the package and replaced with indirect calls (the reason for the short time between releases). (#1129)

• Updates to the way footnotes are printed in tbl_summary() when there is summary type "continuous2" present. Previously, all footnotes were removed, and now only the "continuous2" footnotes are removed. (#1127)

• Added the continuous variable name/label to the footnote for greater clarity in tbl_continuous() (#1123)

• Now exporting the .create_gtsummary_object() function as a utility for other packages to build gtsummary tables. (#1130)

• Fix when add_overall() was run before add_n(). The overall row was not being omitted from the sum and the Ns were doubled. (#1126)

• Added method "emmeans" to add_difference() for tbl_svysummary() objects.

• Updated default add_difference() for tbl_svysummary() objects to be "emmeans" for continuous and dichotomous variables, and "smd" for categorical variables.

gtsummary 1.5.1

New Functions

• Added new function modify_column_alignment() to updated column alignment. Function is a wrapper for the more complex modify_table_styling() function.

• New function tbl_strata2() that passes both the the stratified data frame as well as the stratum level to the user function. (#1091)

• Added a add_p.tbl_continuous() method for adding p-values to tbl_continuous() tables. (#1023)

• Added add_overall.tbl_continuous() method. (#1037)

New Functionality

• New test option "emmeans" in add_difference() and add_p() uses the {emmeans} package to estimate marginal means/least-squares means for continuous variables, binary variables and random intercept models. (#1112)

• Column alignment is now recognized in as_kable() and as_kable_extra(). Previously, the alignment utilized the kable() defaults and ignored any alignment instructions included in the gtsummary table styling.

• The as_kable_extra() was updated to utilize kableExtra::column_spec() and kableExtra::cell_spec() to apply bold and italic styling. The choice of the function depends on the use of escape= in knitr::kable() (#1107)

• Default arguments to knitr::kable() may now be overwritten by passing ... to either as_kable() or as_kable_extra(). Previously, passing a user-defined argument previously in use would result in error.

• Added tbl_strata(.header=) argument providing greater control over the stratum headers that are added to the tables, e.g. you can now add Ns to the headers using this argument.

• Added add_p.tbl_cross(test.args=) argument. (#1095)

• The tbl_strata(.combine_args=) has been added that lets you control all arguments in the tbl_merge() or tbl_stack() that occurs in tbl_strata(). (#1090)

• Added the add_ci(pattern=) argument, which makes it easier to merge the CI column with the primary statistics column. (#1029)

• Suppress tbl_merge() spanning headers by passing tbl_merge(tab_spanner = FALSE) (#1067)

• Functions as_tibble(), as_kable(), and as_kable_extra() gain the fmt_missing= argument that applies missing symbols to missing values. The as_tibble() argument defaults to FALSE, while the others' default is TRUE. (#1073)

• Adding tbl_regression(conf.int=) and tbl_uvregression(conf.int=) argument. For some models, the confidence interval adds to the computation time significantly and may not be needed. This argument will omit the CI calculation. (#1052)

• The add_stat() function was updated to accept tbl_continuous() tables.

• Multinomial models computed using MICE are now supported. (#1065)

• Added theme element to control the tbl_regression(conf.int=) default argument.

• It is now possible to pass a single tbl to tbl_merge(). This is useful when using tbl_merge() as a helper in other functions. (#1068)

• Added statistics= and digits= arguments to the add_overall() family of functions. (#1047)

• Added digits= argument to tbl_cross(). (#1046)

Other Updates

• Allowed modification to font_size for compact theme. (#1106)

• Automatically reduced vertical white space between columns for compact flextable theme.

• Improved user interface for modify_*() functions (#1064)

• Improved error messaging throughout the package. (#1050)

• Added link to the syntax help file to functions throughout the package. The syntax help file illustrates how to use the gtsummary selectors and details the formula-list notation. (#981)

• Updated Spanish translation for Wilcoxon Rank-sum Test.

• Updates and additions to Portuguese language translations. (#1098)

• Updates to the French translations.

• Updating the add_overall() S3 method to have a more common structure, e.g. add_overall(x, ....), where previously, the ... were not present. (#1066)

• Updated the theme_gtsummary_journal("qjecon") to set tbl_regression(conf.int = FALSE) by default.

• Improved error messaging in tbl_custom_summary()

• Updated the default formatting functions in tbl_custom_summary(). Previously, summaries with character results erred because the default summary function was style_number(). This has been updated to style_sigfig() for numeric columns, and as.character() for everything else. (#983)

• All style_*() functions will retain attribute, such as the names. (#1035, #1031, #981)

• The add_n.tbl_regression() (which is also utilized in tbl_uvregression()) was adding the N column without applying a formatting function. The style_number() function has now been added as the default styler. (#1022)

• Added class "tbl_continuous" to the output of tbl_continuous().

• Adding add_p() test "mcnemar.test.wide" to calculate the p-value when the data are stored in a wide format, e.g. one column for a before value and a second column for after. The other McNemar test variant available in {gtsummary} expects data in a long format.

• Converted tbl_split() to S3 function.

• Update how calls to gt::fmt_missing() are constructed to be more memory efficient.

Bug Fixes

• Fix in add_significance_stars() that led to an error when the summarized model did not have a confidence interval column.

• Fix in as_flex_table() and as_hux_table() where reference row was not properly placed after a tbl_merge() when the merged tables share common categorical variables but different reference rows. (#1063)

• Fix in inline_text.gtsummary() where the first level of a categorical variable could not be selected if the table had also been processed with remove_row_type(). (#1078)

• Fix in modify_table_styling(cols_merge_pattern) when it is used with tbl_stack() followed by tbl_merge(). (#1057)

• Bug fix in separate_p_footnotes() where test names were not being translated when theme_gtsummary_language() was set. (#1055)

• Fix in tbl_merge() when rows in a merging table are not present in the first table. (#1033)

• Fix in as_tibble() for nnet::multinom() regression models.

• Fix in tbl_continuous() when include= is not specified.

• Fix in tbl_regression() when a tidier returns CI columns that are all missing. (#1012)

• Fix in add_p()/add_difference() when check whether the passed test is an internal method or a custom method. The previous code required Suggested packages, such as, {lme4}, {effectsize}, and {survey}, to be installed. (#1018)

Breaking Changes

• No longer exporting assert_package(). It has been migrated to {broom.helpers} and we now use broom.helpers::.assert_package(). (#1051)

gtsummary 1.5.0

New Functions

• Added new function tbl_continuous() to summarize a continuous variable by 1 or more categorical variables.

• Added new function add_ci() that adds a new column with the confidence interval for proportions/means reported in tbl_summary(). (#868)

• Migrated a new function tbl_split() from the {bstfun} package. Function allows users to split a {gtsummary} table into multiple tables.

• Migrated a new function separate_p_footnotes() from the {bstfun} package. Function allows users to separate the composite footnote listing the tests performed in add_p(), and replaces it with individual footnotes for each test name.

• New function tbl_custom_summary() allowing to create a table of summary statistics using a custom summary function (#973, #976)

  • Set of helpers to be used with tbl_custom_summary(): continuous_summary(), proportion_summary(), ratio_summary()

• New function modify_cols_merge() that can merge two or more columns in a {gtsummary} table. (#939)

• Added function tbl_butcher() to reduce the size of a {gtsummary} table. After an object has been butchered, other {gtsummary} functions may not be able to execute on the object.

• Added new function tidy_robust() that will add robust standard errors, confidence intervals, and p-values with tbl_regression() and tbl_uvregression(). The function is a wrapper for parameters::model_paramters(). (#979)

New Functionality

• Added a CITATION file so users can now cite the R Journal manuscript using citation("gtsummary").

• Added Standardized Mean Difference method to add_difference(), wrapping the {smd} package's calculations. (#966)

• Extended add_difference() to accept tbl_svysummary() objects in addition to tbl_summary() objects.

• Added a standardized mean difference method for tbl_svysummary() tables.

• Added tbl_strata(.stack_group_header=) argument to include/exclude the headers when tables are combined with tbl_stack()

• Added tbl_strata(.quiet=) argument.

• Allow add_p() and add_difference() to be run on the same table. (#959)

• Updated add_overall() to include the overall statistics in the df_stats tibbles saved in .$meta_data. This makes it possible to report any of the overall statistics using the inline_text(pattern=) argument.

Other Updates

• Added a help file detailing the formula list notation used throughout the {gtsummary} package. (#981)

• Updates to tbl_regression() documentation. The model N is no longer reported by default, and removed that section from the help file. (#998)

• Updates to make the internal df_stats objects consistent across various {gtsummary} objects. Added internal function df_stats_to_table_body that adds the numeric df_stats tibble to .$table_body. The formatting functions are also added for the new columns to .$table_styling$fmt_fun. This function is now used in inline_text.gtsummary() to prepare the returned statistics (#921)

• Now using broom::tidy() to prepare the car::Anova() results. This will be more stable than the version originally written. (#898)

• The function assert_package() now takes the minimum required version of the package from the DESCRIPTION file, and the function is now exported.

• Now using broom::tidy() to prepare aov() test results in add_p.tbl_summary(), which adds additional columns to .$table_body() (#956)

• Updated the README to include links to a recording of a gtsummary presentation and to the RStudio Education blog post.

• Removed maturing lifecycle tag from README.

• Updated deprecated function workflows::pull_workflow_fit(x) to workflows::extract_fit_parsnip(x).

Bug Fixes

• Fix in tbl_summary() when a factor variable is passed that is all NA with no specified levels. (#977)

• Fix in add_p.tbl_summary() when a factor variable with all NA values is passed. (#977)

• Bug fix for the add_difference(estimate_fun=) argument.

Breaking Changes

• Updated add_p.tbl_summary(test = . ~ "mcnemar.test", group = id) syntax to require the group= argument to align with the paired Wilcoxon rank-sum test and paired t-test syntax.

• Deleted deprecated functions add_comparison(), add_global(), tab_style_bold_p(), tab_style_bold_labels(), tab_style_italicize_levels(), tab_style_italicize_labels(), tab_style_bold_levels().

• The following deprecated arguments have been removed: tbl_summary(group=), as_gt(omit=).

• The survival package has been moved from Imports to Suggests, and will no longer automatically be installed when {gtsummary} is installed. Additionally, survival::Surv() is no longer re-exported with the package.

gtsummary 1.4.2

• Update to the internals of tbl_stack() to better handle when two or more stacked tables are then stacked again (#906)

• Updates to make tbl_svysummary() compatible with {survey} package updates in v4.1 (#930)

• The as_hux_table() function previously stripped markdown old syntax from column headers and spanning headers. The output now uses markdown syntax in the headers by default utilizing huxtable::set_markdown() (#885)

• Variables passed in the tbl_svysummary(by=) argument will now automatically be added to include=. (#925)

• Bold and italic requests are now ignored for kableExtra output. These are carried out via markdown syntax, which is not supported by {kableExtra} (#917)

• Bug fix for add_p.tbl_cross(pvalue_fun=); argument was being ignored.

• Updated style_pvalue() to format p-values slightly larger than 1 and slightly lower than 0 (due to imprecise numeric storage). (#907)

• Fix allowing for factor vectors to be passed in tbl_stack(group_header=). (#908)

• Updated arguments y= and x= in tbl_uvregression() to allow for non-standard names (#912)

gtsummary 1.4.1

• Updated tbl_regression() to be compatible with models created with the {parsnip} and {workflows} packages (#890, #647)

• Added the modify_table_styling(text_format = "indent2") option to double indent a row. (#864)

• Messaging update when inline_text.gtsummary() suspects a variable has been repeated in the gtsummary table. (#855)

• Bug fix in add_p.tbl_summary() for columns that are all NA. These variables no longer error; rather, a message is printed indicating the p-value is not possible to calculate. (#889)

• Updated tbl_svysummary() to be compatible with {srvyr} package (#886)

• Updated default header when using tbl_uvregression(x=) to "**Outcome**" (#867)

• The tbl_summary(by=) variable is now added to include= by default (#871)

• Variables are converted to numeric before being passed to wilcox.test() in add_p(). This avoids an error when a date difference is passed. (#880)

• Bug fix for {Hmisc} labeled data with tbl_summary() (#876)

• Bug fix in add_n.tbl_summary() to proportion missing in some cases. (#903)

• Updates to the default formatting functions in the add_glance_*() functions. P-values are now styled with style_pvalue() and No. Obs. and degrees of freedom with style_number() (#870)

gtsummary 1.4.0

New Functions

• Added new function add_glance_table() as a companion to add_glance_source_note(). Function adds model statistics, such as R-squared, to the bottom of the model summary table.

• Added new function add_significance_stars() adding star indicators to significant estimates and an explanatory footnote.

• Added new function tbl_strata(). The function aids prepares gtsummary tables stratified by one or more variables (#679)

• Adding coefficient plot() methods for tbl_regression() and tbl_uvregression(). Function creates a forest plot of model coefficients via GGally::ggcoef_plot().

• New function modify_fmt_fun() has been introduced to help update the functions that format numeric columns and rows in .x$table_body.

• New function introduced, modify_table_styling(), to update printing instructions of tables. This function replaces modify_table_header(), which is now soft deprecated.

• Added function add_vif() to include variance inflation factors in tbl_regression() output. (#717)

• Added a generic function inline_text.gtsummary() that can report results from any gtsummary table. (#398)

New Functionality

• Print infrastructure has been updated to allow for both row and column specification when formatting data or table styling. The x$table_header object has been replaced with a more general x$table_styling. Review the updated vignette "gtsummary_definition.Rmd" for details. The x$table_body is no longer grouped after tbl_stack(); rather, the grouping variable is specified in gt::gt(groupname_col=)

• tbl_summary() now accepts any class as input. Previously only non-date base classes were accepted. For non-base R classes, the summary type must be specified using tbl_summary(type=). The default summary statistic for dates/times is the minimum and maximum. (#488)

• The add_stat() function may now return multiple columns of new statistics. Some arguments have been deprecated in the update. (#746)

• tbl_uvregression() now accepts both data frames and survey design objects as input. (#742)

• If the default tidy_fun = broom::tidy fails, parameters::model_parameters() attempts to tidy the model, if {parameters} is installed. (#854)

• Added a custom tidier for mgcv::gam() models (tidy_gam()) and a method function (tbl_regression.gam()) that uses the new tidier by default. (#745)

• Added default support for brmsfit model in tbl_regression() with new method function. (#751)

• Functions modify_footnote() and modify_spanning_header() now include the text_interpret= argument indicating whether to use gt::md() or gt::html() to format text. Default is gt::md().

• Added/updated functions add_n() and add_nevent() to work with tbl_regression and tbl_uvregression objects. Each function now has an argument to place Ns on the label or level rows. (#744)

• Added The Quarterly Journal of Economics to theme_gtsummary_journal(). This journal theme will be updated again after the gt package updates cols_merge() with a rows argument and allows for line breaks within cell.

• Korean and Icelandic language translations added for theme_gtsummary_language().

• Ability to merge two or more columns with modify_table_styling(cols_merge_pattern=) argument.

• Added theme element "pkgwide-fun:pre_conversion". The function specified here will be executed on the gtsummary object before it is printed or converted with the as_gt(), as_flex_table(), etc functions. (#773)

• The modify_table_body(fun=) argument has been generalized to accept formula shortcut notation.

• Added exploratory data analysis theme that shows more details by default, theme_gtsummary_eda().

Other Updates

• Updated tbl_merge() and tbl_stack() to capture the first source note and caption in the merged/stacked tables. Previously, any captions and source notes were lost.

• Added messaging when table caption requested via modify_caption() for a gt table when using a version of gt that does not support captions.

• Updates to the table gallery vignette reflecting changes in the package. (#775)

• Improved error messaging when there is a problem constructing one of the univariate regression models in tbl_uvregression().

• Added variable-specific formatting to add_difference(estimate_fun=) allowing a single table to show, for example, mean and rate differences that are formatted/rounded differently.

• Improved handling and messaging to users when columns with "haven_labelled" class are passed to gtsummary functions. (#805)

• Improved handling of ordered factors as the by= variable (#569, #540)

• Removed {usethis} package dependency and replaced with {cli}. (#768)

• Added the survey-adapted t-test to theme_gtsummary_mean_sd() for continuous variables in add_p.tbl_svysummary() (#758)

• Allowing for tidyverse shortcut notation in tbl_survfit(estimate_fun=) specification, e.g. tbl_survfit(estimate_fun= ~style_sigfig(.x * 100)) (#761)

• The JAMA journal theme has been updated to merge the coefficient and confidence interval columns.

• Updated other inline_text() functions to wrap inline_text.gtsummary()

Bug Fixes

• The label= argument for unstratified models was being ignored in tbl_survfit() (#842)

• Preserve ordering for factor variables in tbl_survfit(). (#764)

• Bug fix for spanning headers with kableExtra output. The spanning header was misplaced when the header text was sandwiched between two blank spanning headers.

• Bug fix when displaying an unobserved level of a factor variable dichotomously in tbl_summary(). (#780)

• Bug fix in add_p() where test name footnote was not being translated with theme_gtsummary_language().

Breaking Changes

• tbl_survival() has moved from questioning to deprecated. This function maintains the old x$table_header format (instead of the more flexible x$table_styling). The "level_label" column was renamed to "groupname_col" and the x$table_body is no longer grouped with group_by(level_label)

• The back-end implementation of add_stat_label(location = "row") has been updated. The function now merges columns "label" and "stat_label" instead of modifying the "label" column directly. This could be a breaking change if users had manually updated results, for example, from a tbl_regression() table to merge with tbl_summary() using tbl_merge()

• The function add_stat_label() no longer auto-switches location = "column" requests to "row" in the presence of "continuous2" variables.

gtsummary 1.3.7

• No changes. Resubmitting to resolve CRAN quirk.

gtsummary 1.3.6

New Functions

• Added function add_difference(), which adds difference between groups, confidence interval and p-value. (#617)

• Added function modify_caption() that includes table captions. For gt output, requires gt version > 0.2.2 (#701)

• Added function modify_table_body() allowing users to more easily make changes to gtsummary tables

• Added function remove_row_type() for removing header, reference, or missing rows from a gtsummary tables. (#678)

• Added selecting function all_stat_cols() that selects columns from tbl_summary/tbl_svysummary object with summary statistics (i.e. "stat_0", "stat_1", "stat_2", etc.).

• New selecting function was added all_tests() to make it easier to select variables based on the test used to calculate the p-value, e.g. add_p(test = c(age, marker) ~ "t.test", test.args = all_tests("t.test") ~ list(var.equal = TRUE))

• Added functions modify_column_hide() and modify_column_unhide() to hide and unhide columns in .$table_body. Simple wrappers for modify_table_header().

New Functionality

• Previously, the tbl_summary(digits=) only accepted integer values specifying the number of decimal places to round a statistic. The argument now accepts both integers or functions, e.g. digits = age ~ style_sigfig. (#708)

• The add_stat() function now supports tbl_svysummary() objects in addition to tbl_summary() (#688)

• The add_stat() function can now place statistics on the label row, or the level rows for categorical variables. (#714)

• inline_text.tbl_survfit() updated to allow users to select p-value (and other) columns. (#589)

Other Updates

• modify_header() has been updated so users may more easily access internal data while defining headers and so that it no longer adds its call to the gtsummary .$call_list (#719)

• The default value for include= argument for the add_global_p.tbl_regression() and add_global_p.tbl_uvregression() methods have been made the same with include = everything(), and the help files for these methods have been brought together in a single file. (#721)

• Introducing new package dependency {broom.helpers}

  • The tidying and preparation of tbl_regression() tables are now being performed by {broom.helpers} (#636, #607)

  • All select helper functions and the utility functions that make them possible, have been cleaned up and migrated to {broom.helpers}. (#648, #680)

  • Importing .generic_selector(), .select_to_varnames(), and .formula_list_to_named_list() from {broom.helpers}: this is a function that makes it easy to create selecting functions like all_continuous(). The internals allow for it to be used in {broom.helpers} and {gtsummary} seamlessly. (#680)

  • Theme element has been added for controlling the other tidy_plus_plus() arguments. (#692)

  • Variables that do not follow standard naming conventions are now parsed correctly and presented in the regression tables.

• The tbl_regression() function is now an S3 method. The new interface allows for special handling of different model types using S3 methods: tbl_regression.default(), tbl_regression.lmer(), tbl_regression.glmer(), tbl_regression.survreg()

• tbl_regression(add_estimate_to_reference_rows=) argument has been added. Also added to tbl_uvregression(). (#692)

• Theme element for tbl_regression(add_estimate_to_reference_rows=) has been added. (#677)

• Removed "Statistics presented:" and "Statistical tests performed:" prefixes from the tbl_summary() %>% add_p() footnotes.

• The codebase powering add_p() and related methods has been refactored for better performance, organization, and customizability. (#622)

  • For clarity, a help file listing each test available within gtsummary and the pseudo code for calculating the p-value has been added (see ?tests)

  • Each add_p() method now has the test.args= argument. Use this argument to pass additional arguments to the statistical method, e.g. add_p(test = c(age, marker) ~ "t.test", test.args = c(age, marker) ~ list(var.equal = TRUE))

  • Additional tests have been added: paired t-test, signed rank test, and more. See ?tests for full list.

  • More robust unit testing implemented for all add_p() methods.

• Added messaging to tbl_stack() to inform users that the attributes from the first table passed take precedent over the others'. (#699)

• In the modfiy_*() functions, if users did not select any columns, they encountered an error. Now, if no columns are selected, instructions are printed for how to correctly select columns in a gtsummary table. Moreover, if no columns are selected, the gtsummary object is now returned unaltered. (#699)

• The add_glance_source_note() function has been generalized so users may pass any glance function. Previously, broom::glance() was being used with no option to change it. (#699)

• The ... arguments have been added to as_gt(). These dots are subsequently passed to gt::gt(...). (#701)

• Multiple imputation models created with {mice}, and multinomial regression models created with {nnet} are now supported in tbl_regression() (#645)

• Updates to add_global_p.tbl_regression() allowing for variable names with spaces and special characters (#682)

• Added digits= argument to style_percent() (#690)

• Users may now choose which tbl_regression() columns to report with a theme element. they can choose among the "estimate", "std.error", "statistic", "ci", "conf.low", "conf.high" and "p.value" (#637)

• Allow users to include the reference value in tbl_regression() via a theme element

• Users may change the symbol with a reference row symbol with a theme element. (#628)

Bug Fixes

• Bug fix when a default statistic is set using themes for "continuous2" variables that has length larger than one

• Bug fix when missing/non-missing percentages requested in add_n.tbl_summary()

Breaking Changes

• The default test for 2-by-2 tables with expected cell counts has been updated from the chi-squared test with continuity correction to the original chi-squared test for add_p.tbl_summary() (#721)

• Experimental function add_p.tbl_survfit(test.args=) in addition to accepting the formula list notation, also accepted a single string naming a test that was interpreted as everything() ~ "test_name". The single string is no longer accepted, and users must use the formula notation.

• Removed theme element N_fun that was previously marked as questioning and likely to be removed from the package.