Multivariable Model - Building: A Pragmatic Approach to Regression Anaylsis based on Fractional Polynomials for Modelling Continuous Variables

Multivariable Model - Building: A Pragmatic Approach to Regression Analysis Based on Fractional Polynomials for Modelling Continuous Variables by Patrick Royston, and Willi Sauerbrei is a textbook appropriate for clinical trialist and researchers in biomedical science, who are engaged in a daily basis in building multivariable prognostic models. While a number of texts exist for these audiences, this textbook was written at a more accessible level for those without any advanced statistical knowledge. The book begins laying the groundwork for model building as well as underlying the importance of modeling the appropriate functional form of continuous variables in order to minimize any residual confounding. Moreover, the authors have impressive coverage of the different statistical methodologies available for evaluating the functional form of continuous variables, like fractional polynomials and restrictive cubic splines. The advantages and drawbacks for each of these methods are also provided. The authors make use of statistical software packages throughout to demonstrate many of statistical procedures it presents. STATA is the primary package used and many of the necessary commands are given for readers who wish to take advantage of software to assist them in analyzing data. In addition to STATA, the authors provided the ink for resources in SAS and R. For survival analysis, STATA is the primary package used with details given on how to implement the methods covered in the text. This is an amazing book! As a physician-epidemiologist, I strongly recommend it. If you want to excel in multivariable model building, you will not be disappointed.