19756

Video URL

https://simons.berkeley.edu/talks/Learning-from-RCTs-in-Public-Health-and-Medicine

Learning from RCTs in Public Health and Medicine

(2022). Learning from RCTs in Public Health and Medicine. The Simons Institute for the Theory of Computing. https://simons.berkeley.edu/talks/Learning-from-RCTs-in-Public-Health-and-Medicine

Learning from RCTs in Public Health and Medicine. The Simons Institute for the Theory of Computing, Feb. 14, 2022, https://simons.berkeley.edu/talks/Learning-from-RCTs-in-Public-Health-and-Medicine 

          @misc{ scivideos_19756,
            doi = {},
            url = {https://simons.berkeley.edu/talks/Learning-from-RCTs-in-Public-Health-and-Medicine},
            author = {},
            keywords = {},
            language = {en},
            title = {Learning from RCTs in Public Health and Medicine},
            publisher = {The Simons Institute for the Theory of Computing},
            year = {2022},
            month = {feb},
            note = {19756 see, \url{https://scivideos.org/index.php/Simons-Institute/19756}}
          }
James Robins (Harvard University)
Talk number19756
Source RepositorySimons Institute
Subject

Abstract

RCTs are potentially useful in many ways other than standard confirmatory intent to treat (ITT) analyses, but to succeed difficult problems must be overcome.I will discuss some or (time-permitting) all of the following problems : 1. The problem of transportability of the trial results to other populations: I will explain why transportability is much more difficult in trials comparing longitudinal dynamic treatment regimes rather than in simple point treatment trials. 2. The problematic use of RCT data in micro-simulation models used in cost-benefit analyses 3.The problem of combining data from large, often confounded, administrative or electronic medical records , with data from smaller underpowered randomized trials in estimating individualized treatment strategies. 4. The problem of using the results of RCTs to benchmark the ability of observational analyses to 'get it right', with the goal of providing evidence that causal analyses of observational data are sufficiently reliable to contribute to decision making 5.The problem noncompliance with assigned protocol in trials in which the per-protocol effect rather than the ITT effect is of substantive importance . 6. The problem of leveraging the prior knowledge that diagnostic tests have "no direct effect on the outcome except through the treatment delivered" to greatly increase the power of trials designed to estimate the cost vs benefit of competing testing strategies.