Calculating sample size for studies with expected all-or-none nonadherence and selection bias.

Area of Science:

• Biostatistics
• Clinical Trial Design
• Epidemiology

Background:

• Nonadherence (noncompliance) and selection bias can complicate clinical trial analysis.
• Previous work addressed increased variances due to nonadherence but assumed no selection bias.

Purpose of the Study:

• To develop sample size formulas for studies with expected nonadherence and selection bias.
• To extend existing methods to incorporate selection bias as systematic differences in latent adherence subgroups.
• To provide tools for accurate sample size calculations in clinical trials.

Main Methods:

• Developed sample size formulas for mean differences between treatment and control groups.
• Extended prior work on nonadherence variances to include selection bias.
• Incorporated systematic differences in means and variances among adherence subgroups.
• Utilized pilot adherence data for sample size calculations.

Main Results:

• New formulas are presented for sample size calculations under nonadherence and selection bias.
• The approach accounts for increased variances and systematic differences.
• Illustrative calculations are provided for planning clinical trials.
• Formulas for normally distributed outcomes and uncertainty from pilot data are available.

Conclusions:

• The developed formulas enhance the accuracy of sample size determination in complex clinical trial scenarios.
• Accounting for both nonadherence and selection bias is crucial for robust study planning.
• The methods offer practical tools for researchers designing clinical trials.