National Cancer Institute Home at the National Institutes of Health |
Please wait while this form is being loaded....
The Applied Research Program Web site is no longer maintained. ARP's former staff have moved to the new Healthcare Delivery Research Program, the Behavioral Research Program, or the Epidemiology & Genomics Research Program, and the content from this Web site is being moved to one of those sites as appropriate. Please update your links and bookmarks!

Publication Abstract

Authors: Gold HT, Sorbero ME, Griggs JJ, Do HT, Dick AW

Title: Structural estimates of treatment effects on outcomes using retrospective data: an application to ductal carcinoma in situ.

Journal: Med Care Res Rev 68(6):627-49

Date: 2011 Dec

Abstract: Analysis of observational cohort data is subject to bias from unobservable risk selection. The authors compared econometric models and treatment effectiveness estimates using the linked Surveillance, Epidemiology, and End Results (SEER)-Medicare claims data for women diagnosed with ductal carcinoma in situ. Treatment effectiveness estimates for mastectomy and breast-conserving surgery (BCS) with or without radiotherapy were compared using three different models: simultaneous-equations model, discrete-time survival model with unobserved heterogeneity (frailty), and proportional hazards model. Overall trends in disease-free survival (DFS), or time to first subsequent breast event, by treatment are similar regardless of the model, with mastectomy yielding the highest DFS over 8 years of follow-up, followed by BCS with radiotherapy, and then BCS alone. Absolute rates and direction of bias varied substantially by treatment strategy. DFS was underestimated by single-equation and frailty models compared with the simultaneous-equations model and randomized controlled trial results for BCS with radiotherapy and overestimated for BCS alone.

Last Modified: 03 Sep 2013