Software | Multivariable Fractional Polynomials (MFP)

STATA

FP, MFP and all the extensions have been programmed by Patrick Royston in Stata. FP (commands fracpoly and fp) and MFP (command mfp) have been ‘factory’ routines, i.e. standard parts of Stata, for many years.

The following articles have appeared in the Stata Journal.

R&S (2007): Multivariable modelling with cubic regression splines: A principled approach.
R&S (2009): Bootstrap assessment of the stability of multivariable models.
R&S (2009): Two techniques for investigating interactions between treatment and continuous covariates in clinical trials.
Royston (2014): A smooth covariate rank transformation for use in regression models with a sigmoid dose-response function.
R&S (2015) mfpa: extension of mfp using the acd covariate transformation for enhanced parametric multivariable modelling.

In addition, the FP- or MFP-related packages listed below are available for download within Stata from Royston's UCL website via the following Stata command:

net from http://www.homepages.ucl.ac.uk/~ucakjpr/stata/
fp_plus: Extensions to mfp and fracpoly supporting factor variables
fracpoly_powers: Extensions to mfp and fracpoly supporting factor variables
metacurve: Meta analysis of a continuous covariate
mfpboot: Bootstrapping MFP models for stability analysis
mfpi: Modelling interactions between categorical and continuous variables
mfpigen: Modelling interactions between continuous variables
mfpmi: MFP for multiply imputed data
stpmfp: Flexible parametric survival modelling with fractional polynomials
stmfpt: Multivariable Cox models with time-dependent covariate effects

MFP in R

The MFP in R software is implemented in two packages: mfp (Ambler G, and Benner A, 2023) and mfp2 (Kipruto E, Kammer M, Royston P, Sauerbrei W, 2023). The main advantage of mfp2 relative to mfp is informative output, computational efficiency, the ability to model sigmoid functions, availability of plotting functions and an additional interface that takes response y and design matrix X. The output of mfp2 is similar to mfp in Stata.

MFP in SAS

MFP is also available as a SAS macro.

For more details see Sauerbrei et al (2006): Multivariable regression model building by using fractional polynomials: description of SAS, STATA and R programs.