• When a binary endpoint interrupts another binary endpoint
• LJ Wei: see Claggett 2012
• Brown dissertation: https://era.library.ualberta.ca/items/03a9ab6c-2882-45ed-948e-f2568ac8f899/view/26b0a5fe-1cb3-4721-b24c-60ca6454567c/Brown_Paul_M_201705_PhD.pdf
• FH philosophy: Multiplicity adjustment for testing multiple endpoints or subsets is necessary when the investigator wishes to elevate the endpoint or subset yielding an impressive result to prime focus, replacing other questions. It is not necessary when a sequence of tests is pre-specified and all results are reported in this pre-specified context.
• Multiple ordinal longitudinal outcomes with GEE and correlation structure for both within- and between endpoints: jia17joi
• https://www.the-scientist.com/?articles.view/articleNo/46348/title/Composite-Endpoints-in-Clinical-Trials/
• https://www.mjcrowther.co.uk/pdf/Crowther_megenreg_2017.pdf
• https://twitter.com/ihtanboga/status/1034166300573097984
• https://onlinelibrary.wiley.com/doi/pdf/10.1002/pst.1909
• Rafiqul I Chowdhury and M Ataharul Islam have a paper under review 2019-02 at PLOS ONE: Risk prediction of a sequence of ordinal outcomes from repeated measures: regressive modeling approach using proportional, partial proportional odds and multinomial logistic regression models. Uses strictly regressive models (earlier outcomes index parameters for later periods, plus baseline covariate adjustment). Doesn’t apply to treatment comparisons but more for one-sample prediction/trajectory analysis. I was a reviewer for the second round.
• ICH E9 addendum especially this