# Modeling Long-term Outcomes and Treatment Effects After Androgen Deprivation Therapy for Prostate Cancer

**Authors:** Yolanda Hagar, James J. Dignam, Vanja Dukic  
**arXiv:** [1509.01275](https://arxiv.org/abs/1509.01275)  
**Date:** 2015-09-03  
**Summary page:** https://www.summarizepaper.com/en/paper/1509.01275v1/

## Key points

- Authors explore complexities of analyzing outcomes in long-term cancer survivor studies
- Compare seven different survival models that estimate hazard rate and consider proportional and non-proportional covariates
- Focus on extension of multi-resolution hazard (MRH) estimator with non-proportional hierarchical MRH approach and data-driven pruning algorithm
- Approach allows for computational efficiency and robust estimates with few observed failures
- Use data from large-scale randomized prostate cancer clinical trial to examine patterns of biochemical failure and estimate time-varying effects of treatment and other covariates
- Comparing modeling strategies and smoothness assumptions to assess impact on estimated treatment effect
- Results show benefits of treatment diminish over time, informing future research and clinical decision making
- Study provides insights into modeling long-term outcomes in cancer survivor studies, emphasizing need to account for non-proportional hazards when assessing treatment effects.

## Summary

In the study "Modeling Long-term Outcomes and Treatment Effects After Androgen Deprivation Therapy for Prostate Cancer," authors Yolanda Hagar, James J. Dignam, and Vanja Dukic explore the complexities of analyzing outcomes in long-term cancer survivor studies. To address these challenges, they compare seven different survival models that estimate the hazard rate and consider both proportional and non-proportional covariates. Specifically, they focus on an extension of the multi-resolution hazard (MRH) estimator which combines a non-proportional hierarchical MRH approach with a data-driven pruning algorithm. This approach allows for computational efficiency and produces robust estimates even when there are few observed failures. Using data from a large-scale randomized prostate cancer clinical trial, the researchers examine patterns of biochemical failure and estimate the time-varying effects of androgen deprivation therapy treatment and other covariates. By comparing different modeling strategies and smoothness assumptions, they assess how these choices impact the estimated treatment effect. The results reveal that the benefits of treatment diminish over time which may inform future research and clinical decision making in this field. Overall, this study provides valuable insights into modeling long-term outcomes in cancer survivor studies and highlights the need to account for non-proportional hazards when assessing treatment effects.