The full modern playbook on this way of thinking doesn't exist yet. The estimand way of thinking is still fairly new, developed during the last 10 years. It is on the frontiers of pharmaceutical statistics and much debate is ongoing on how granular it needs to be even beyond the ICH guidelines. Most of the complexity comes from illnesses where RCTs in practice are quite pragmatic — as in cancer where you don't want to deny patients access to better treatments or need to account for patient drop-outs for reasons related to the illness and so forth. Clinical trials in psychiatry do not usually have events as outcomes unlike oncology but they have many complicated reasons for missing outcomes. The estimand framework has barely come to psychiatry though. Oncology has a working group with educational material and reviews and it is slowly coming up in other areas like neurology.
Makes sense. Thanks so much for your input. I think we certainly need something like this for psychiatry as results across RCTs, at times, can be 1. Perplexingly inconsistent and 2. Poorly predictive of real-world efficacy and tolerability…
Thanks— do you have a good resource on good clinical trial design hygiene?
The full modern playbook on this way of thinking doesn't exist yet. The estimand way of thinking is still fairly new, developed during the last 10 years. It is on the frontiers of pharmaceutical statistics and much debate is ongoing on how granular it needs to be even beyond the ICH guidelines. Most of the complexity comes from illnesses where RCTs in practice are quite pragmatic — as in cancer where you don't want to deny patients access to better treatments or need to account for patient drop-outs for reasons related to the illness and so forth. Clinical trials in psychiatry do not usually have events as outcomes unlike oncology but they have many complicated reasons for missing outcomes. The estimand framework has barely come to psychiatry though. Oncology has a working group with educational material and reviews and it is slowly coming up in other areas like neurology.
Makes sense. Thanks so much for your input. I think we certainly need something like this for psychiatry as results across RCTs, at times, can be 1. Perplexingly inconsistent and 2. Poorly predictive of real-world efficacy and tolerability…