Examination Date
Fall 8-4-2025
Degree
Thesis
Degree Program
Master of Science (MS) Biostatistics
Examination Committee
Dr. Zhide Fang, Dr. Hui-Yi Lin, Dr. Anand Paul
Abstract
Prostate cancer remains one of the most common and deadly cancers among men, with significant disparities in survival outcomes based on race, socioeconomic status, and treatment access. While five-year survival rates appear high, they often mask long-term differences, especially among patients with advanced disease or limited care options. In this study, overall survival (OS) measured from diagnosis to death from any cause or last follow-up per the SEER definition was used as the primary outcome. SEER data from 9,664 prostate cancer cases (2000–2021) were analyzed to examine how clinical and sociodemographic factors influence survival, and to compare the performance of four survival models: Kaplan-Meier, Cox Proportional Hazards, Weibull, and Exponential.Summary statistics revealed key disparities in age, PSA levels, Gleason scores, and treatment patterns. Kaplan-Meier curves showed significant survival differences by age, race, Gleason score, and PSA level. The Cox model identified age, PSA, Gleason score, marital status, and income as strong predictors of survival, while surgery showed a modest survival benefit. Among racial groups, Non-Hispanic Asian/Pacific Islander patients had the best outcomes. Parametric models confirmed many of these findings, though the Exponential model underperformed due to its assumption of a constant hazard rate. Model comparison using AIC indicated that the Cox model provided the best overall fit, followed by the Weibull model.These findings highlight the importance of selecting the right survival models in cancer research and make it clear that survival depends on both medical and social factors. To improve prostate cancer outcomes, especially for patients with limited access to care, accurate risk prediction and fair treatment options are needed.
Recommended Citation
Maldonado, Adriana L., "Comparative Analysis of Survival Models in Prostate Cancer Using SEER Data" (2025). School of Public Health. 14.
https://digitalscholar.lsuhsc.edu/etd_sph/14
A. Maldonado signature page