Shortening the baseline period in hot flash symptom control trials: Comparison of a one-day versus seven-day baseline hot flash diary.

Abstract

e24094 Background: Hundreds, if not thousands, of participants are enrolled hot flash trials each year given the magnitude of the problem for breast and prostate cancer survivors. Classically, these studies have included a baseline period of 1-week, or longer, to understand the frequency and severity of hot flashes before starting a new treatment. The objective of the current study was to compare the accuracy of a 1-day baseline hot flash diary to a more traditional 1-week dairy. A shorter baseline period would be more convenient for patients and provide a more efficient study design. Methods: Raw data from 5 pilot studies and 15 phase III randomized, controlled trials (RCTs), all of which used a one-week baseline period, were obtained. Descriptive statistics were used to describe day-by-day variations in hot flash frequencies and scores, during the baseline week. Additional analyses evaluated whether the conclusions from any of the individual pilot studies would have been changed if only a 1-day baseline period had been used. For the RCTs, p-values were recalculated using mixed models, adjusting for the baseline value by including it as a covariate. Results: A total of 2,573 participants were included. On average, participants had 8.5 hot flashes per day on Day 1. Mean hot flash frequencies and scores on subsequent days (Days 2-7) were within 6% of Day 1 values. When comparing a 1-day to a 1-week baseline period, there was an absolute difference of only 0.29 hot flashes per day (SD 2.25). Re-analysis for each pilot study revealed that no individual study conclusions would have been meaningfully altered by a shorter baseline. For the RCTs, a shorter baseline period changed the results of only 1 of 24 comparisons from statistically significant to not significant, or vice versa. Conclusions: A 1-day hot flash diary appears to accurately reflect true frequency and severity of baseline hot flash symptoms in appropriately sized cohorts. Results support a shorter baseline period for pilot studies and RCTs.