A Modified Additive Hazard Model for Some Risk Factors Associated with Hypertensive Condition

Abstract:

In clinical studies, the effect of covariate in their various forms is most often difficult to interpret via the proportional hazard assumption of the cox regression. In this study, proportional hazard assumption test was carried out to access the compliance of some risk factors associated with hypertensive condition (HTC). The global test indicates a violation to proportional hazard assumption. Therefore, a Bayesian survival model was proposed with generalized additive predictors, and modified for the hypertensive dataset for which continuous and linear covariate forms were incorporated in a proportional hazard framework. The findings of the study are: for the Risk Factors (overweight and pregnancy), both categorical variables failed the test with p-values 0.00017 and 0.00030 respectively, less than the significant level 0.05; results suggest that a Regularized Hypertensive Additive Hazard model (RHAHM) performs better with smaller Deviance Information Criterion (DIC) value of 2933.323 and large Log-Peusdo Maximum Likelihood (LPML) value of 1469.415 in comparison to the un-regularized model. Also, age covariate at the seventh knot are more at risk of High-Risk Hypertensive Condition (HRHC) than those of other knots, followed by those who enter the study at the beginning with HRHC. Male patients are about 10% less at risk of HRHC than their female counterparts; those that are married are 59% more at risk of HRHC than subjects that are single. For factors which could be a possible cause of HRHC, patients with chronic illness and those who indulge in excessive alcohol intake have the highest risk of HRHC than those of other risk factors. The study concludes that regular medical check-up on hypertensive status will effectively help in the management of hypertension, especially those with target organ damage.