Handling Missing Data in Ambulatory Blood Pressure Monitoring Safety Studies: A Simulation Study of Common Analytical Methods

Ambulatory blood pressure monitoring studies are recommended as important safety studies in the 2022 Assessment of Pressor Effects of Drugs FDA Guidance for Industry. To facilitate implementation, we investigated missing data issues in ABPM studies and recommend appropriate statistical methods. Seven missing data methods were evaluated using simulated data with missing proportions ranging from 2% to 50% and sample sizes ranging from 100 to 500, under single-arm and two-arm (with placebo) study designs.Method performance in controlling Type I error varied by missing data rate and study design. In single-arm studies, methods performed comparably at lower missing rates; however, multiple imputation applied to 24-hour average data is preferred, as it effectively controls Type I error without being overly conservative. AIPW or IPW outperformed other methods when the missing rate exceeded 25%. For two-arm studies with placebo, methods performed similarly.This project provides important insights for statistical analysis of missing data in ABPM studies. MI is recommended as the primary analytical approach. Sensitivity analyses using alternative methods such as AIPW or IPW are also recommended, especially in scenarios with high missing data rates. When a missing not at random mechanism is suspected, we suggest tipping point analysis as a sensitivity analysis.