Sample size formulae for intervention studies with the cluster as unit of randomization.

Area of Science:

• Biostatistics
• Clinical Trials
• Epidemiology

Background:

• Cluster randomization is frequently used in intervention studies.
• Determining appropriate sample size is crucial for study power and validity.
• Existing sample size methods may not adequately address the complexities of cluster randomization.

Purpose of the Study:

• To present novel sample size formulae for cluster randomized intervention studies.
• To provide guidance on determining the number of clusters or individuals per cluster.
• To facilitate the design of studies with continuous or dichotomous endpoints.

Main Methods:

• Derivation of sample size formulae based on Student's t-test.
• Inclusion of both within and between cluster variance components.
• Development of power contours for study design.
• Separate formulae for stratified and unstratified cluster designs.

Main Results:

• Formulae for calculating sample size when either the number of clusters or individuals per cluster is known.
• Presentation of power contours to aid in sample size determination.
• Demonstration of how to account for within- and between-cluster variance.

Conclusions:

• The presented formulae offer a robust method for sample size calculation in cluster randomized trials.
• These tools can improve the efficiency and statistical power of intervention studies using cluster randomization.
• The findings support better planning and execution of public health and medical intervention studies.