What statistical test is most appropriate for analyzing the mean change in clinic-based systolic blood pressure (BP) from baseline to day 14 between acetaminophen (paracetamol) and placebo treatment periods in a crossover trial?

Statistical Analysis for PATH-BP Trial: Comparing Mean Change in Clinic-Based Systolic BP

The paired t-test is the most appropriate statistical test for analyzing the mean change from baseline in clinic-based systolic BP between acetaminophen and placebo treatment periods in the PATH-BP crossover trial. 1

Rationale for Using Paired T-Test

The paired t-test is the correct choice for this analysis based on several key factors:

1. Study Design Characteristics:

   • Crossover design where each participant receives both treatments (acetaminophen and placebo)
   • Paired measurements from the same individuals
   • Continuous outcome variable (change in systolic BP)
   • Interest in comparing means between two treatment conditions

2. Data Structure in PATH-BP Trial:

   • Each participant has two measurements (baseline and day 14) for each treatment period
   • The primary analysis compares the mean change from baseline between treatment periods
   • The data shows normally distributed continuous variables (BP measurements)

Why Other Tests Are Not Appropriate

• McNemar test: Used for paired nominal data (comparing proportions in matched pairs), not for continuous variables like blood pressure 2

• Chi-square test: Appropriate for unpaired categorical data, not for continuous paired measurements 2

• Wilcoxon rank sum test: Used for unpaired data or when normality assumptions are violated; the paired equivalent would be Wilcoxon signed-rank test 2

Statistical Analysis Approach in Similar Studies

In comparable clinical trials examining blood pressure changes:

• The American College of Cardiology/American Heart Association guidelines recommend paired t-tests for analyzing continuous outcomes in crossover trials where the same participants receive multiple treatments 2

• The European Urology Association suggests using standard deviation (SD) to calculate effect sizes in clinical trials, which aligns with the paired t-test approach used in the PATH-BP trial 1

Implementation Details

When implementing the paired t-test for this analysis:

1. Calculate the change from baseline for each participant in both treatment periods
2. Compute the difference between these changes for each participant
3. Perform a paired t-test on these differences
4. Report the mean difference, 95% confidence interval, and p-value

This approach accounts for within-subject correlation and provides the most statistical power for detecting treatment effects in this crossover design.

Common Pitfalls to Avoid

• Failing to account for the paired nature of the data would substantially reduce statistical power
• Using unpaired tests would incorrectly treat the observations as independent
• Not accounting for period effects in crossover trials can bias results (though the washout period in PATH-BP helps mitigate this)

The paired t-test is specifically designed for this type of analysis and is the standard approach recommended by statistical guidelines for analyzing continuous outcomes in crossover trials.