Original Article

Comparison between fractional polynomials, spline smoothing, and multiple logistic regression models in the study of associated hypertension risk factors


Background & Aim: Previous studies about hypertension and risk factors have shown the linear relationship between them. However, we can improve the fit of models with some changes and have a better form for estimation of coefficients and interpret the effects of variables.
Methods & Materials: This survey was a cross-sectional study from 2010 to 2011 in Yazd, Iran. The participants were among the subjects aged from 40 to 80. Body mass index (BMI), sex, age, renal failure, history of diabetes (years of disease), type of diabetes (type 1 or type 2), the number of cigarettes per day and years of smoking were predictors and the binary response returned to hypertension (yes or no). The traditional logistic model was used for determining the relationship between covariates and the outcome. Then, the models were modified with multivariable fractional polynomials.
Results: Our findings displayed fitting the multivariable fractional polynomials (MFP) model in the parametric model which was the best fit for the modeling. The difference deviance in MFP was 21.952 (P < 0.001). The linear model in comparison with null model deviance differences was 22.170 (P < 0.001). The second-degree fractional polynomials model compared with first-degree fractional polynomials model, and the difference deviance was 21.850 (P < 0.001).
Conclusion: MFP model approach is an alternative procedure that can solve previous problems about the categorical approach, step function, and cut- off points.

World Health Organization. A global brief on hypertension: Silent killer, global public health crisis. Geneva, Switzerland: WHO; 2013.

Royston P, Sauerbrei W. A new approach to modelling interactions between treatment and continuous covariates in clinical trials by using fractional polynomials. Stat Med 2004; 23(16): 2509-25.

Wong ES, Wang BC, Garrison LP, Alfonso-Cristancho R, Flum DR, Arterburn DE, et al. Examining the BMI-mortality relationship using fractional polynomials. BMC Med Res Methodol 2011; 11: 175.

Royston P, Ambler G, Sauerbrei W. The use of fractional polynomials to model continuous risk variables in epidemiology. Int J Epidemiol 1999; 28(5): 964-74.

World Health Organization. Hypertension [Online]. [cited 2013]. Available from: URL: http://www.who.int/topics/hypertension/en.

World Health Organization. Obesity and overweight [Online]. [cited 2011]; Available from: URL:


Medline Plus. Kidney failure [Online]. [cited 2012]; Available from: URL:


World Health Organization. Archives of WHO [Online]. [cited 2014]; Available from: URL: http://www.who.int/archives/en.

Blair M. Diabetes Mellitus Review. Urol Nurs 2016; 36(1): 27-36.

Royston P, Altman DG. Regression using fractional polynomials of continuous covariates: Parsimonious parametric modelling. J R Stat Soc 1994; 43(3): 429-67.

Elbashir M, Wang J, Wu FX, Wang L. Predicting beta-turns in proteins using support vector machines with fractional polynomials. Proteome Sci 2013; 11(Suppl 1): S5.

Royston P. Multivariable regression models with continuous covariates, with a practical emphasis on fractional polynomials and applications in clinical epidemiology. Konstanz, Germany: German Stata Users' Group Meetings; 2005.

Binder H, Sauerbrei W, Royston P. Comparison between splines and fractional polynomials for multivariable model building with continuous covariates: A simulation study with continuous response. Stat Med 2013; 32(13): 2262-77.

Benner A. Multivariable fractional polynomials [Online]. [cited 2005]; Available from: URL:


Royston P, Sauerbrei W. Building multivariable regression models with continuous covariates in clinical epidemiology-with an emphasis on fractional polynomials. Methods Inf Med 2005; 44(4): 561-71.

Cui J, de Klerk N, Abramson M, Del MA, Benke G, Dennekamp M, et al. Fractional polynomials and model selection in generalized estimating equations analysis, with an application to a longitudinal epidemiologic study in Australia. Am J Epidemiol 2009; 169(1): 113-21.

Judd KL. Numerical methods in economics. Cambridge, MA: MIT ress;

Faes C, Aerts M, Geys H, Molenberghs G. Model averaging using fractional polynomials to estimate a safe level of exposure. Risk Anal 2007; 27(1): 111-23.

Price AMH, Quach J, Wake M, Bittman M, Hiscock H. Cross-sectional sleep thresholds for optimal health and well-being in Australian 4-9-year-olds. Sleep Med 2016; 22: 83-90.

Schmidt CO, Ittermann T, Schulz A, Grabe HJ, Baumeister SE. Linear, nonlinear or categorical: How to treat complexassociations in regression analyses? Polynomial transformations and fractional polynomials. Int J Public Health 2013; 58(1): 157-60.

Greenland S. Dose-response and trend analysis in epidemiology: Alternatives to categorical analysis. Epidemiology 1995; 6(4): 356-65.

Zhang Z. A mathematical model for predicting glucose levels in critically-ill patients: The PIGnOLI model. PeerJ 2015; 3: e1005

IssueVol 3 No 2 (2017) QRcode
SectionOriginal Article(s)
Likelihood functions Logistic regression model Statistical models Hypertension Body mass index

Rights and permissions
Creative Commons License This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
How to Cite
Ganji M, Yekaninejad MS, Yaseri M. Comparison between fractional polynomials, spline smoothing, and multiple logistic regression models in the study of associated hypertension risk factors. JBE. 2018;3(2):49-59.