The “story time” is to lull us in with a randomized controlled experiment and as we fall asleep, feed us less reliable conclusions that come from an embedded observational study.

Kaiser Fung explains. This comes up a lot, and his formulation in the above title is a good way of putting it.

He also has this discussion of the AstraZeneca-Oxford vaccine trial results which makes me want to just do a damn Bayesian analysis of it already. I’ll have to find someone with the right combination of inclination and subject-matter expertise.

7 thoughts on “The “story time” is to lull us in with a randomized controlled experiment and as we fall asleep, feed us less reliable conclusions that come from an embedded observational study.

  1. I started reading the discussion on the AO vaccine trial when the following snippet about what changed in the study mid-course appeared: “a) The vaccine arm was swtiched from one dose to two doses” And the first thing that came to mind was, why would they switch from left arm to right arm depending on how many doses you get? And that is why I am just a lurker here.

  2. Despite the messed up AO experiment design, I still think we have enough evidence to say it’s safe, and at least 50% effective in younger people. Couldn’t the FDA just approve it for younger people? We need all the effective vaccines we can get right now.

    • The FDA has decided to wait until AztraZeneca completes their Phase III trials here in the US, and I’m fine with that, myself.

      It’s not clear how many doses they could supply to the US at the moment, anyway.

      • The whole situation is a bit farcical. They claimed the accidental half first dose followed by a full second dose had the best VE (90%). If true, that complicates the operation massively – the doses are no longer interchangeble! Now, they appear to be arguing that one dose is enough. Presumably that would be a single full dose. But why not a single half dose? And each one of these analyses have different underlying analysis sets so they can’t really be compared directly.

    • I don’t know, every dollar we spend on AO vaccines is a dollar we don’t spend on a nearly twice as effective vaccine right? (Pfizer or Moderna). It seems better to put money towards expanding Pfizer and Moderna’s production rate, and maybe optimizing dosing. I mean, you might be able to **double** the available Pfizer and Moderna vaccines if you dose at half as much, and it still might be **more** effective than 50% right? (of course it might not)

      Also, we’re going to want modified vaccines in the next few months due to variants. RNA vaccines are nearly trivial to modify compared to adenovirus or other techniques.

Leave a Reply

Your email address will not be published. Required fields are marked *