A simple correction to completer analyses and improvement on baseline observation carried forward

Area of Science:

• Biostatistics
• Clinical Trial Methodology
• Data Analysis

Background:

• The baseline observation carried forward (BOCF) and last observation carried forward (LOCF) methods are frequently used for handling missing data in research despite known issues.
• These methods can lead to biased results and are often recommended against in statistical guidelines.

Purpose of the Study:

• To address the persistent use of problematic missing data techniques like BOCF and LOCF.
• To develop and present a simple, statistically sound method for adjusting completer analyses when dealing with missing follow-up data.
• To provide practical tools for reanalyzing existing studies, adjusting power calculations, and assessing the plausibility of missing data handling.

Main Methods:

• Demonstrated the equivalence of BOCF in single-sample change testing to a completer analysis.
• Derived a summary-statistics-based method to adjust completer analysis inference.
• Assumed zero mean change from baseline for missing data, with variance inflation based on completer data.
• Extended the method for two-sample comparisons.

Main Results:

• The proposed method provides an adjustment to completer analysis, effectively acting as a BOCF with variance inflation.
• Simulations were used to explore the properties of the new method.
• The method was applied to a weight loss program trial dataset.

Conclusions:

• The developed method offers a practical approach to re-evaluating studies with inadequate missing data handling.
• It serves as a valuable tool for power calculations and plausibility checks in clinical research.
• This method complements existing robust techniques for managing missing data, promoting more accurate and reliable research findings.