Amazing summary (including the VA) and discussion of a complicated story. Congrats! #NephJC

Sometimes more lessons are learned from failures than successes. I've learned quite a bit from working through the challenges of this trial. #NephJC

I would also add that the separation of 0.8 mg/dl is inflated by the baseline difference in phosphate (i.e., not all the result of the intervention). Also, the outcomes of all-cause mortality & hospitalization have substantial "noise." #NephJC

It was all randomized... however, the cluster-randomized portion of enrollment was likely biased. #NephJC

Correct. Electronic consent with educational materials and videos was mostly successful. The lack of dedicated coordinators- especially on-site coordinators- really left us vulnerable to events such as the pandemic and workforce turnover. #NephJC

The pivot to individual randomization was successful with regard to the reasons for the pivot: baseline phosphate narrowed substantially. Enrollment is a different story. #NephJC

Important to highlight the timing here. Enrollment was held for much of 2020 when the pandemic hit. The equally large (and lingering) challenge was the burden placed on the dialysis staff in the wake of the pandemic. This affects all aspects of trial without research staff. #NephJC

Enrollment? Yes. Opt-out consent would have led to incredible enrollment (see: TiME). However, this does not translate into fidelity of the intervention/phos separation (see: TiME). #NephJC

There is some hesitation to sign up for something you've been told is harmful... also hesitation on the part of the dialysis facility staff to approach select patients for this reason. #NephJC

This was not the intended design. The pivot to individual randomization was made to avoid the issues encountered with cluster randomization in this trial (namely post-randomization consent biasing enrollment). #NephJC

Correct- this (presumably) led to post-randomization bias. This was evident by difference in baseline phosphate (6.6 vs 5.7 mg/dl) and consent rates (38% vs 63%) for high vs low, respectively. #NephJC

Important design element here- HiLo obtained informed consent. Compare this to TiME, which was an opt-out consent for data collection (not the intervention). #NephJC

I also want to highlight that the HiLo PI was Myles Wolf (now at Cornell), and the trial leadership included Laura Dember (Penn), Tamara Isakova (NW), and Hrishikesh Chakraborty (DCRI). #NephJC