Bayesian Joint Modeling of Longitudinal and Survival Time Measurement of Hypertension Patients

Abstract

Background

High blood pressure is a health risk for all populations, worldwide. Globally the number of people with uncontrolled hypertension rose by 70% between 1980 and 2008.

Objective

This paper aims to investigate the association of survival time and fasting blood sugar levels of hypertension patients and identify the risk factors that affect the survival time of the patient.

Methods

We considered a total of 430 random samples of hypertension patients who were followed-up at Yekatit-12 Hospital in Ethiopia from January 2013 to January 2019. A linear mixed effects model was used for the longitudinal outcomes (fasting blood sugar) with normality assumption, although four parametric accelerated failure time distributions: exponential, Weibull, lognormal and loglogistic are studied for the time-to-event data. The Bayesian joint models were defined through latent variables and association parameters and with specified noninformative prior distributions for the model parameters. Simulations are conducted using Gibbs sampler algorithm implemented in the WinBUGS software. The model selection criteria DIC is employed to identify the model with best fit to the data.

Results

The findings from Bayesian joint models are consistent. The association parameter in each Bayesian joint model is significant. This implies that there is dependence between the two processes: longitudinal fasting blood sugar level and the time-to-death event under joint models. With investigation of the model comparison criteria, the Bayesian–Weibull model was preferred to analysize the current data sets. Based on joint analysis the baseline age, place of residence, family history of hypertension, khat intake, blood cholesterol level of the patient, hypertension disease stage, adherence to the treatment and related disease were associated factors that affect the survival time of hypertension patients.

Conclusion

The analysis suggests that there is strong association between longitudinal process (fasting blood sugar) and time-to-event data. The researcher recommends that all stakeholders should be aware of the consequences of these factors which can influence the survival time of hypertension patients in the study area.

Keywords:

• Bayesian
• joint model
• hypertension
• survival analysis
• parametric models

Acknowledgments

The author would like to sincerely thank the Yeketit12 Hospitals for providing the data sets used in this study. The anonymous reviewers are acknowledged for their detailed comments and suggestions.

Ethical Consideration

The ethical clearance was checked and approved by ethical clearance committee of Arba Minch University Department of Statistics and the Addis Ababa Administration Health Bureau Yekatit 12 Hospital Medical College medical director’s office granted permission to use the patients’ data for this study. For the purpose of confidentiality, there were no links with individual patients and all data had no personal identifier and were kept confidential and therefore did not require informed consent.

Disclosure

The author declares no conflicts of interest regarding the publication of this paper.