Sample size calculation for randomized trials via inverse probability of response weighting when outcome data are...

Area of Science:

• Biostatistics
• Clinical Trials Methodology
• Epidemiology

Background:

• Randomized trials are essential for evaluating intervention effectiveness.
• Missing outcome data is a common challenge in clinical trials.
• Current methods for sample size calculation with missing data may be inadequate, particularly with informative missingness.

Purpose of the Study:

• To investigate and propose a robust method for sample size calculation in randomized trials with missing outcome data.
• To address the limitations of standard sample size adjustments when missingness is informative.
• To provide practical tools for researchers conducting trials with missing data.

Main Methods:

• Utilizing an inverse probability of response weighted (IPRW) estimating equations approach.
• Applying M-estimation theory to derive sample size formulas.
• Developing formulas for both individually randomized and cluster randomized trials (CRTs).

Main Results:

• Derived novel sample size formulas accounting for missing at random outcome data under IPRW.
• Demonstrated the application of the method through a sample size calculation for a HIV testing strategies CRT.
• Developed an R shiny app to facilitate the implementation of the proposed sample size formulas.

Conclusions:

• The proposed IPRW approach offers a more accurate method for sample size calculation in the presence of missing outcome data.
• The derived formulas and accompanying R shiny app provide valuable resources for trial design.
• This work enhances the reliability of sample size estimations in clinical trials with potential missing data.