Statistically Adjusting for Wear Time in Randomized Trials of Continuous Glucose Monitors as a Complement to Intent-To-Treat and As-Treated Analyses: Application and Evaluation in Two Trials.