2 updates on a previous post on trial terminations - First, I did look at the data by phase. Maybe not shockingly, phase 3 trials had a clearly lower termination rate (just under 14% - though that's still higher than I would have thought). The "slash" trials, phase 1/2 and phase 2/3, were considerably higher, presumably with a number of those terminating at the phase transition, per plan. Second - I tried parsing the data by trial size (target enrolled), but realized that the data is not reliable: some sponsors retain the original estimated enrollment, but others update to the actual enrolled at termination. This is actually the MORE interesting result to me, as it illustrates how these kinds of analyses can go sideways very easily if you're not looking very skeptically at your data*. (*Obligatory Feynman quote: "The first principle is that you must not fool yourself and you are the easiest person to fool.")
Bruce Pendleton
5mo
Much appreciated. Talk soon
Patient-Centered CGO @ Centricity Research
5moThe fact the actual enrollment can overwrite estimated enrollment is one of the biggest challenges of this data set! I have had to use Python to pull the records once they change to recruitment status to collect estimated recruitment before it gets overwritten. Monthly and daily snapshots from CTTI AACT 👨🏼🍳 🐍 📊.