A bonus following my #RedemptionWeek:
I’m highlighting a short paper that can be used for teaching caution in #ClinicalTrial subgroups:
“‘Thursday’s child has far to go’—interpreting subgroups and the #STAMPEDEtrial” https://www.annalsofoncology.org/article/S0923-7534(19)34965-8/fulltext
1/6
Using real #ClinicalTrial data, our example allows readers to think about the plausibility of subgroup analyses.
This was solicited by Annals of Oncology (via Gert Attard) after Nick James presented at ASCO 2017. Mel Spears was 1st author.
2/6
A key lesson is to specify upfront both the planned analysis and the expected direction of effect. Anything else should be noted as “exploratory”.
The full version of Figure 1 is worth exploring (abridged version here).
3/6
I raise this paper here in hope people will use this, if they aren’t already, for #Teaching on their MSc in Clinical Trials courses , such as that run by UCL's Institute of Clinical Trials and Methodology (https://www.ucl.ac.uk/prospective-students/graduate/taught-degrees/clinical-trials-msc)
4/6
There are other good examples in the literature, including that from ISIS-2 which tongue-in-cheek-ly uses the spoon-bendy nonsense of astrology as it's jumping off point. We went more mundane.
5/6
I hope my set of #RedemptionWeek threads on #ClinicalTrials and #Methodology has been in some way useful to a couple of people. It’s certainly been cathartic for me.
6/6
Matt Sydes