Conditional power for assessing population interventions.

Area of Science:

• Biostatistics
• Clinical Trial Design
• Pharmaceutical Research

Background:

• Comparative two-period studies often involve analyzing data collected over time.
• Accurate power calculations are crucial for designing effective clinical trials.
• Existing power formulae may not fully account for previously observed baseline data.

Purpose of the Study:

• To develop a method for calculating conditional power in comparative two-period studies.
• To specifically address studies with previously observed baseline data.
• To modify standard power formulae by isolating variability from future data.

Main Methods:

• Developed a novel approach to conditional power calculation.
• Modified standard power formulae to incorporate observed baseline data.
• Isolated variability attributable to yet-to-be-observed data.

Main Results:

• Demonstrated the method using opioid overdose rates before and after product reformulation.
• Conditional power was influenced by comparator population size and assumed secular changes.
• The study examined null, alternative, and ancillary hypotheses related to the reformulation impact.

Conclusions:

• Pre-post study designs can be initiated after baseline data collection.
• Conditional power calculations based on observed baseline data offer a distinct approach to variability.
• This method provides a more nuanced understanding of statistical power in longitudinal studies.