We were interested to read the paper by Dimitriadis et al.1 that was published in Hypertension Research in June 2016. The authors aimed to evaluate the predictability power of body mass index (BMI), waist circumference (WC) and waist-to-hip ratio (WHR) for the incidence of coronary artery disease (CAD) in a cohort of essential hypertensive patients. The result of the Cox proportional hazard model showed that, among the studied anthropometric indices, WC can only be a predictor of the future development of CAD.1
Although the data were interesting, some methodological and statistical issues should be considered. It is unclear how the predictors of the final Cox regression model are determined in the study. In addition, it seems that some clinically sensible interactions among the variables were missed in the study. Investigators typically use the step-wise methods for model building. First, bivariate correlations among variables are assessed to test potential multicollinearity. Second, they conducted the regression model with backward step-wise selection for those variables that were found to be associated with the studied outcome on univariate analysis (P-value <0.2), as well as for clinically sensible interaction terms.2
As the authors note in their conclusion, WC can only predict the future development of CAD; such a conclusion is an optimistic interpretation unless the prediction model can be validated internally or externally.3
Moreover, we suggest that the authors can reanalyze their data to evaluate the power of each of the studied predictors to discriminate which subjects will develop CAD using the C-statistics or area under the Receiver Operating Characteristics (ROC) curve (AUC) and then to examine the statistical difference between the AUCs using statistical methods such as Hanely and McNeil or DeLong.4, 5
The concluding message for the readers is that, for modern clinical prediction, several powerful and valuable tools are available for model building and risk prediction.
References
Dimitriadis K, Tsioufis C, Mazaraki A, Liatakis I, Koutra E, Kordalis A, Kasiakogias A, Flessas D, Tentolouris N, Tousoulis D . Waist circumference compared with other obesity parameters as determinants of coronary artery disease in essential hypertension: a 6-year follow-up study. Hypertens Res 2016; 39: 475–479.
Hosmer DW Jr, Lemeshow S . Applied Logistic Regression. John Wiley & Sons. 2004.
Steyerberg E . Clinical Prediction Models: A Practical Approach to Development, Validation, and Updating. Springer Science: & Business Media: New York, NY, USA. 2008.
DeLong ER, DeLong DM, Clarke-Pearson DL . Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988; 44: 837–845.
Hanley JA, McNeil BJ . The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982; 143: 29–36.
Acknowledgements
The authors thank statistics consultants at the Research Development Center of Sina Hospital for their technical assistance.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Competing interests
The authors declare no conflict of interest.
Rights and permissions
About this article
Cite this article
Ayubi, E., Sani, M., Khazaei, S. et al. Waist circumference compared with other obesity parameters as determinants of coronary artery disease in essential hypertension: statistical issues to avoid misinterpretation. Hypertens Res 40, 516 (2017). https://doi.org/10.1038/hr.2016.168
Published:
Issue Date:
DOI: https://doi.org/10.1038/hr.2016.168