3 ms·
Randomizing people to have sex with an HIV+ or placebo partner would bring down the IRB's wrath quite quickly. For that reason, a lot of this stuff is done obse
by mattkrause 8y ago
Randomizing people to have sex with an HIV+ or placebo partner would bring down the IRB's wrath quite quickly. For that reason, a lot of this stuff is done observationally, where you just track people doing whatever they were going to do anyway. The data is often not as clean as a randomized controlled trial, but it seems much more ethical.
An earlier trial (HPTN 052) did actually run a RCT, but the treatment strategy was randomized, not the partner. It worked so well that the trial was actually stopped early and everyone was offered antivirals. Cohen et al. (2011) reported the results here: https://www.nejm.org/doi/10.1056/NEJMoa1105243?url_ver=Z39.88-2003 https://www.nejm.org/doi/10.1056/NEJMoa1105243?url_ver=Z39.8...
- Fomite 8y agoYou can get near to an RCT with newer trial designs like stepped-wedge designs, which are way better in terms of equipoise and nearly as efficient statistically.
- mattkrause 8y agoThose are still interventional though, right? (I probably should have written observational vs. interventional instead of obs. vs RCT).
- Fomite 8y agoYep. The stepped-wedge design, for example, is one where everyone will get the intervention, but the time in which they get it is randomized.