[R-pkgs] Updates to Hmisc, rms, and greport packages

Frank Harrell f.harrell at Vanderbilt.Edu
Thu Apr 17 02:01:53 CEST 2014

The Hmisc package has had a number of updates/fixes:

Changes in version 3.14-4 (2014-04-13)
    * rcspline.eval: stop new logic for ignoring outer values when there 
are many ties when there are also many ties on interior values.  Added 
new logic to use interior unique values of x when the number of unique x 
is small.
    * latexBuild: generalized with insert argument
    * latex.default: modified to use mods of latexBuild, fixed bug with 
caption.loc='bottom' (thanks: YacineH)
    * latex.default: fixed bug where comma did not appear after 
caption={} for ctable (thanks: Johannes Hofrichter)
    * tests: fit.mult.impute.bootstrap.r: added new example (thanks: 
Jane Cook)
    * fit.mult.impute: added call for fit.mult.impute in returned 
object, replacing call from fitter; makes update work for fit.mult.impute
    * summary.formula: fixed recognition of multiple left-hand-side 
variables to trigger call to summaryM (thanks: Kevin Thorpe)
    * summaryM: changed group label to '' instead of 0 for formulas like 
age + sex ~ 1
    * Ecdf: added what argument to all functions
    * nobsY: return constructed id vector
    * addMarginal: instead of .marginal. being logical, make it contain 
names of variables being marginalized over
    * mChoice.c: fixed some inconsistencies

The rms package has also had some changes, and the survfit.formula 
function has been REMOVED, replaced by the npsurv function:

Changes in version 4.2-0 (2014-04-13)
    * Deprecated survfit.formula so would not overlap with function in 
    * Added function npsurv, survplot.npsurv
    * REMOVED survfit.formula
    * Used new type argument to label.Surv for fitting functions
    * cph: added weights argument to residuals.coxph (Thanks: Thomas Lumley)
    * survfit.cph: fixed bug in using wrong type variable.  Thanks: 
Zhiyuan Sun
    * cph: added weighted=TRUE in call to residuals.coxph (Thanks: T Lumley)
    * orm.fit: improved ormfit to not try to deal with NaN in V, 
assuming that step-halving will happen

Some improvements have been made in the greport package:

Changes in version 0.5-1 (2014-04-15)
    * survReport: changed to use npsurv instead of survfit.formula
	 * exReport: changed order of output so that analysis of randomized 
patients marked for exclusions appears last; use LaTeX chngpage package 
to allow detailed table to go into left margin so as to be centered on page
    * exReport: added adjustwidth argument
	 * accrualReport: allowed enrollment target N to be omitted
    * exReport: fine tuning
    * nriskReport: new report to show number of subjects still being 
followed at each day
	 * Merge: added support for data.table
    * nriskReport: added id() variable
    * exReport: fixed bug when there is an exclusion with 0 frequency
    * accrualReport: improved graphics formatting, added minrand argument
    * accrualReport: added enrollmax argument, clarified notation
    * exReport: added ignoreExcl, ignoreRand arguments
    * all: added greportOption texwhere; default is 'gentex'; can 
specify texwhere='' to write non-appendix LaTeX code to console as for knitr
    * dReport: for byx for discrete Y, sense when Y is binary and use 
Wilson interval instead of bootstrap; adjust SE using confidence 
interval if proportion is 0 or 1
    * dReport: changed discreteness non-binary classification to use 
maximum number of unique values or Y instead of minimum
    * add globalVariables call to nriskReport

Frank E Harrell Jr Professor and Chairman      School of Medicine
                    Department of Biostatistics Vanderbilt University

More information about the R-packages mailing list