Intention-to-treat analysis when only a baseline value is available

Area of Science:

• Biostatistics
• Clinical Trials
• Data Analysis

Background:

• Intention to treat (ITT) analysis is crucial for clinical trial integrity.
• Handling missing follow-up data, especially with baseline values but no subsequent measurements, presents analytical challenges.

Purpose of the Study:

• To compare the performance of different methods for intention to treat (ITT) analysis when patients lack follow-up measurements.
• Evaluate no imputation (standard and alternative mixed models), single imputation (baseline value carried forward), and multiple imputation (selective and non-selective) techniques.

Main Methods:

• A simulation study was conducted with varying scenarios: missingness association with baseline values, treatment receipt status, and percentage of missing data.
• Two real-life datasets were analyzed to validate simulation findings.

Main Results:

• Mixed model analyses and multiple imputation demonstrated comparable bias and coverage in most simulated and real-world scenarios.
• Selective imputation, utilizing information on non-receipt of treatment in the allocated treatment group, outperformed other methods in that specific scenario.

Conclusions:

• Standard mixed model analysis without imputation is generally appropriate for ITT analysis with missing data.
• Selective imputation is advised when patients allocated to treatment did not receive it, though results require cautious interpretation.