Title

xmfp-- Multivariable fractional polynomial models with extensions

xmfp[,options]:regression_cmd[yvar1[yvar2]]xvarlist[if] [in] [weight] [,regression_cmd_options]

optionsDescription ------------------------------------------------------------------------- Model 2sequentialuse the Royston and Altman model-selection algorithm; default uses closed-test procedurecycles(#)maximum number of iteration cycles; default iscycles(5)dfdefault(#)default maximum degrees of freedom; default isdfdefault(4)center(cent_list)specification of centering for the independent variablesalpha(alpha_list)p-values for testing between FP models; default isalpha(0.05)df(df_list)degrees of freedom for each predictorpowers(numlist)list of FP powers to use; default ispowers(-2 -1(.5)1 2 3)linadj(varlist)list of variables to be included as untransformed linear terms in all MFP modelsAdv. model

xorder(+|-|n)order of entry into model-selection algorithm; default isxorder(+)select(select_list)nominal p-values for selection on each predictorxpowers(xp_list)FP powers for each predictorzero(varlist)treat nonpositive values of specified predictors as zero when FP is transformedcatzero(varlist)add indicator variable for specified predictorsallinclude out-of-sample observations in generated variablesReporting

level(#)set confidence level; default islevel(95)display_optionscontrol column formats and line width -------------------------------------------------------------------------

regression_cmd_optionsDescription ------------------------------------------------------------------------- Adv. modelregression_cmd_optionsoptions appropriate to the regression command in use -------------------------------------------------------------------------All weight types supported by

regression_cmdare allowed; see weight. See[R] mfp postestimationfor features available after estimation.fracgenmay be used to create new variables containing fractional polynomial powers. See[R] fracpoly.

regression_cmdmay beclogit,glm,intreg,logistic,logit,mlogit,nbreg,ologit,oprobit,poisson,probit,qreg,regress,rreg,stcox,stcrreg,streg, orxtgee.

yvar1is not allowed forstreg,stcrreg, andstcox. For these commands, you must firststsetyour data.

yvar1andyvar2must both be specified whenregression_cmdisintreg.

xvarlisthas elements of typevarlistand/or(varlist); for example,

x1 x2 (x3 x4 x5)Elements enclosed in parentheses are tested jointly for inclusion in the model and are not eligible for fractional polynomial transformation.

xmfpselects the multivariable fractional polynomial (MFP) model that best predicts the outcome variable from the right-hand-side variables inxvarlist.

xmfpprovides some extensions to the factory-standardmfpcommand, namely1.

xmfpsupports factor variables 2.xmfphas alinadj(varlist)option to adjust linearly for variables inva> rlist.Note also that the

mfppost-estimation commandsfracplotandfracpredare replaced withxfracplotandxfracpred, respectively. The syntax is unchanged, except thatxfracplothas an additional optionnoptswhich suppresses plotting of partial residuals.

AuthorPatrick Royston MRC Clinical Trials Unit London, UK pr@ctu.mrc.ac.uk

Also seeManual:

[R] fracpoly,[R] mfpOnline:

mfp