Real-World Effectiveness: Addressing Selection Bias in the Well Elderly 2 Study Strengthens Effect Estimates

Area of Science:

• Gerontology
• Clinical Trials
• Biostatistics

Background:

• Intention-to-treat (ITT) analysis in randomized trials can be limited by participant non-adherence and loss to follow-up.
• These limitations may lead to an underestimation of treatment effects.
• Addressing internal validity threats is crucial for accurate interpretation of trial outcomes.

Purpose of the Study:

• To demonstrate methods for addressing internal validity threats in ITT analyses.
• To apply these methods using data from the Well Elderly 2 study.
• To compare standard ITT with an adjusted ITT approach.

Main Methods:

• Secondary analysis of 4-year data from 460 older adults in the Well Elderly 2 study.
• Utilized inverse probability of censoring weights (IPCW) to adjust for participant follow-up loss.
• Compared outcomes from standard ITT with IPCW-adjusted ITT.

Main Results:

• Standard ITT indicated improvements in 10 out of 17 outcomes.
• IPCW-adjusted ITT demonstrated improvements in 13 out of 17 outcomes.
• For SF-36v2, IPCW-adjusted ITT showed significant improvements in both mental and physical composite scores, unlike standard ITT.

Conclusions:

• IPCW-adjusted ITT analyses may reveal stronger intervention effects compared to standard ITT.
• Standard ITT is more conservative but susceptible to misclassification bias.
• Reporting both standard and adjusted ITT analyses provides a more robust interpretation of treatment effects in the presence of missing data.