1 / 9

Comments on talks

Marshall M. Joffe University of Pennsylvania. Comments on talks. ATS: What’s new?. Already in use previously in trials, practice Consider study: treatment for blood pressure (BP) Protocol specifies changing initial agent if BP too high ATS Compared with standard treatment using ITT

mea
Download Presentation

Comments on talks

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Marshall M. Joffe University of Pennsylvania Comments on talks

  2. ATS: What’s new? • Already in use previously in trials, practice • Consider study: treatment for blood pressure (BP) • Protocol specifies changing initial agent if BP too high • ATS • Compared with standard treatment using ITT • Adaptive part of strategy not focus or evaluated • Standard trials: • Protocols/strategies allow changes due to side effects • Earlier methods not tailored for evaluating second line therapies

  3. Use Standard Design and Analysis • Trial with 4 arms • A: initial treatment • B: second line treatment plan : example • B1: add second treatment if SBP>140 • B2: add second treatment if SBP>160 • Arms: A1B1, A1B2,A2B1, A2B2 • Adaptive treatment strategies • Compare using ITT • Not way to optimize plan

  4. Compare multiple regimes/plans • Tailoring variables • In clinical trials literature for simple treatments, known as subgroup analysis • Extended here to multiple treatments, times, dynamic regimes • Often has exploratory flavor rather than confirmatory • Viewed (in simpler setting) negatively by • Part of clinical trials community • Regulators • Want prespecified comparisons

  5. Issues with multiplicity • Examples: • Just 1 or 2 covariates • In principle: many at each time point • How does one choose covariates/number for tailoring strategies? • Risks: • Overfitting models • Overoptimism of estimates of predictive ability, value of regime • Presenters: • How would one deal with this? • Speak about opportunities and risks involved

  6. Time-varying treatments, covariates • Estimation methods variants of earlier methods developed by Robins • G-formula • G-estimation (not discussed here, less used in other settings; why?) • Inverse probability of treatment weighting • New focus on determining optimal regimes/strategies • Results in non-regularity • Solutions presented

  7. Design • Less discussed • Can randomize • Up front/at beginning to strategy • Sequentially/multiple times • Analysis strategies similar • Compare options

  8. Approaches to selecting covariates for tailoring • Observational studies • Apply similar G-methods • Clinical trials • Wider

  9. Emphasis on pragmatic trials/decisions • Take individual preferences into account? • Maximize individualized utility functions • Earlier work on effects of time-varying treatments had some emphasis on modeling, understanding causal processes • Interest in incorporating here as part of learning systems

More Related