Feed

Survival analysis consists of a set of statistical methods used to analyse data where the outcome variable is the time until an event occurs. When such data are collected across distinct spatial regions, incorporating spatial information into survival models can be beneficial. A common approach is to apply an intrinsic Conditional Autoregressive (CAR) prior to an area-level frailty term to account for spatial correlation between regions. We extend the Bayesian Cox semi-parametric model by incorporating a spatial frailty term using the Leroux CAR prior. The aim was to improve the model's ability to describe stroke hospitalisations at the Stroke Centre Hospital in Makassar, Indonesia with a focus on understanding the geographic distribution of hospitalisations, Length of Stay (LOS) and factors influencing patient outcomes. The dataset was obtained from medical records of stroke patients admitted to this hospital (April 2021-June 2024). Variables included LOS, discharge outcomes, sex, age, stroke type, uric acid levels, hypertension, hypercholesterolemia, and diabetes mellitus. Our findings indicate that diabetes, stroke type and the presence of hypercholesterolemia significantly influence recovery rates in stroke patients. Specifically, patients with diabetes had lower recovery, while those with hypercholesterolemia and ischemic stroke patients had faster recovery compared to those with haemorrhagic strokes.