Quantifying treatment selection bias effect on survival in comparative effectiveness research: findings from low-risk prostate cancer patients.

Miccio JA, Talcott WJ, Jairam V, Park HS, Yu JB, Leapman MS, Johnson SB, King MT, Nguyen PL, Kann BH. Quantifying treatment selection bias effect on survival in comparative effectiveness research: findings from low-risk prostate cancer patients. Prostate Cancer and Prostatic Diseases. 2021 Jun 1; 24(2):414-422.

Related HSR&D Project(s)

• IIR 15-356 – A multi-modal, physician-centered intervention to improve guideline-concordant prostate cancer imaging

Dimensions for VA is a web-based tool available to VA staff that enables detailed searches of published research and research projects.

If you have VA-Intranet access, click here for more information vaww.hsrd.research.va.gov/dimensions/

VA staff not currently on the VA network can access Dimensions by registering for an account using their VA email address.    Search Dimensions for VA for this citation
* Don't have VA-internal network access or a VA email address? Try searching the free-to-the-public version of Dimensions

Search for Abstract from PubMed

---

Abstract:

BACKGROUND: Comparative effectiveness research (CER) using national registries influences cancer clinical trial design, treatment guidelines, and patient management. However, the extent to which treatment selection bias (TSB) affects overall survival (OS) in cancer CER remains poorly defined. We sought to quantify the TSB effect on OS in the setting of low-risk prostate cancer, where 10-year prostate cancer-specific survival (PCSS) approaches 100% regardless of treatment modality. METHODS: The Surveillance, Epidemiology, and End Results database was queried for patients with low-risk prostate cancer (cT1-T2a, PSA? < 10, and Gleason 6) who received radical prostatectomy (RP), brachytherapy (BT), or external beam radiotherapy (EBRT) from 2005 to 2015. The TSB effect was defined as the unadjusted 10-year OS difference between modalities that was not due to differences in PCSS. Propensity score matching was used to estimate the TSB effect on OS due to measured confounders (variables present in the database and associated with OS) and unmeasured confounders. RESULTS: A total of 50,804 patients were included (8845 RP; 18,252 BT; 23,707 EBRT) with a median follow-up of 7.4 years. The 10-year PCSS for the entire cohort was 99%. The 10-year OS was 92.9% for RP, 83.6% for BT, and 76.9% for EBRT (p? < 0.001). OS differences persisted after propensity score matching of RP vs. EBRT (7.4%), RP vs. BT (4.6%), and BT vs. EBRT (3.7%) (all p? < 0.001). The TSB effect on 10-year OS was estimated to be 15.0% for RP vs. EBRT (8.6% measured, 6.4% unmeasured), 8.5% for RP vs. BT (4.8% measured, 3.7% unmeasured), and 6.5% for BT vs. EBRT (3.1% measured, 3.4% unmeasured). CONCLUSIONS: Patients with low-risk prostate cancer selected for RP exhibited large OS differences despite similar PCSS compared to radiotherapy, suggesting OS differences are almost entirely driven by TSB. The quantities of these effects are important to consider when interpreting prostate cancer CER using national registries.