Sample size considerations for matched-pair cluster randomization design with incomplete observations of binary...

Area of Science:

• Biostatistics
• Epidemiology
• Clinical Trials

Background:

• Matched-pair cluster randomized trials (CRTs) are common in public health research.
• Incomplete observations (missing data) pose significant challenges in CRTs.
• Existing methods for sample size determination in CRTs with missing data are inadequate.

Purpose of the Study:

• To propose a novel, closed-form sample size formula for matched-pair CRTs with incomplete binary outcomes.
• To develop a method that accurately accounts for missing data patterns, matching, and clustering effects.
• To provide a more accurate sample size calculation compared to conventional methods.

Main Methods:

• Utilized the generalized estimating equation (GEE) approach within a marginal logistic regression model.
• Developed a closed-form sample size formula addressing various missing data mechanisms and correlation structures.
• Assessed the proposed method's performance through simulation studies.

Main Results:

• The proposed GEE-based sample size method demonstrates higher accuracy than conventional adjustments for missing data.
• The formula effectively incorporates the impact of matching, clustering, and missing data characteristics.
• Simulation studies validated the accuracy and reliability of the new sample size calculation.

Conclusions:

• The developed sample size formula offers a more accurate approach for designing matched-pair CRTs with incomplete binary outcomes.
• This method addresses critical methodological gaps in sample size determination for complex trial designs.
• The approach is applicable to real-world studies, such as evaluating interventions for blood pressure control.