Novel way to evaluate the role of PSA screening by using regression discontinuity design for survival data
Principal Investigator
Name
Youngjoo Cho
Degrees
Ph.D.
Institution
Independent
Position Title
Assistant professor
Email
About this CDAS Project
Study
PLCO
(Learn more about this study)
Project ID
PLCO-379
Initial CDAS Request Approval
Jul 5, 2018
Title
Novel way to evaluate the role of PSA screening by using regression discontinuity design for survival data
Summary
The role of prostate cancer screening remains controversial, partially due to the findings of PLCO prostate cancer trial, which failed to show a reduction in prostate cancer-specific survival or overall mortality by a PSA screening strategy (Andriole et al. 2009). Shoag et al. (2015) suggested that the effect of PSA screening can be alternatively evaluated using a regression discontinuity (RD) method based on the screening arm of PLCO prostate trial only. By exploiting the observation that a PSA of 4.0ng/ml was used as a threshold to prompt additional evaluations, they showed that a 4.0ng/ml PSA threshold provided a statistically significant higher rate of biopsy and a higher detection rate of low-risk prostate cancer. Regression Discontinuity (RD) is an approach has been originally proposed and used widely in econometrics research. In the context of clinical oncology and biostatistics, RD provides a statistical framework to estimate the causal effect of a treatment (e.g. additional diagnosis) that is deterministically determined by a threshold of a continuous baseline variable (e.g., PSA of 4.0ng/ml).
In this proposal, we plan to use RD methods to evaluate the role of PSA screening on prostate cancer-specific survival and overall mortality. Such evaluation relies on state-of-art statistical methodology of RD for both conventional time-to-event data (e.g., overall mortality) and more complicated competing risks data (e.g., prostate cancer-specific survival). To our best knowledge, the development of statistical methodology for RD design in the time-to-event data has been limited. We recently developed a novel estimation and statistical inference approach for the causal treatment effect under nonrandomized studies, where the outcome is time-to-event data and the treatment decision is (at least partially) based on a threshold of a continuous baseline variable. Therefore this methodology is well positioned to more comprehensively elucidate and quantify the role of PSA screening on overall mortality and prostate cancer-specific survival with the threshold.
Reference
Andriole, G.L., Crawford, E.D., Grubb III, R.L., Buys, S.S., Chia, D., Church, T.R., Fouad, M.N., Gelmann, E.P., Kvale, P.A., Reding, D.J., Weissfeld, J.L., Yokochi, L. A. et al. for the PLCO Project Team, (2009). Mortality results from a randomized prostate-cancer screening trial. New England Journal of Medicine, 360(13), 1310-1319.
Shoag, J., Halpern, J., Eisner, B., Lee, R., Mittal, S., Barbieri, C. E. and Shoag, D. (2015). Efficacy of prostate-specific antigen screening: Use of regression discontinuity in the PLCO cancer screening trial. JAMA oncology, 1(7), 984-986.
In this proposal, we plan to use RD methods to evaluate the role of PSA screening on prostate cancer-specific survival and overall mortality. Such evaluation relies on state-of-art statistical methodology of RD for both conventional time-to-event data (e.g., overall mortality) and more complicated competing risks data (e.g., prostate cancer-specific survival). To our best knowledge, the development of statistical methodology for RD design in the time-to-event data has been limited. We recently developed a novel estimation and statistical inference approach for the causal treatment effect under nonrandomized studies, where the outcome is time-to-event data and the treatment decision is (at least partially) based on a threshold of a continuous baseline variable. Therefore this methodology is well positioned to more comprehensively elucidate and quantify the role of PSA screening on overall mortality and prostate cancer-specific survival with the threshold.
Reference
Andriole, G.L., Crawford, E.D., Grubb III, R.L., Buys, S.S., Chia, D., Church, T.R., Fouad, M.N., Gelmann, E.P., Kvale, P.A., Reding, D.J., Weissfeld, J.L., Yokochi, L. A. et al. for the PLCO Project Team, (2009). Mortality results from a randomized prostate-cancer screening trial. New England Journal of Medicine, 360(13), 1310-1319.
Shoag, J., Halpern, J., Eisner, B., Lee, R., Mittal, S., Barbieri, C. E. and Shoag, D. (2015). Efficacy of prostate-specific antigen screening: Use of regression discontinuity in the PLCO cancer screening trial. JAMA oncology, 1(7), 984-986.
Aims
1. Within PLCO prostate cancer trial screening arm, to estimate the causal effect of additional diagnoses that was prompted by observing PSA of 4.0ng/ml on overall survival.
2. Within PLCO prostate cancer trial screening arm to estimate the causal effect of additional diagnoses that was prompted by observing PSA of 4.0ng/ml on cause-specific survival.
Collaborators
Chen Hu, Ph.D., Division of Biostatistics and Bioinformatics, Johns Hopkins University, Baltimore, MD, 21205
Debashis Ghosh, Ph.D., Department of Biostatistics and Informatics, University of Colorado, Aurora, CO, 80045
Herbert Ballentine Carter, M.D., Johns Hopkins School of Medicine, Baltimore, MD, 21205