Should stroke trials adjust functional outcome for baseline prognostic factors?

Area of Science:

• Clinical Trials
• Biostatistics
• Neurology

Background:

• Many stroke clinical trials yield neutral results, potentially due to suboptimal statistical analysis.
• Inefficient analyses may obscure the true effectiveness of interventions.

Purpose of the Study:

• To evaluate if adjusting outcome measures for baseline prognostic factors improves statistical efficiency in stroke trials.
• To quantify the potential sample size reduction achievable through adjusted analyses.

Main Methods:

• Utilized data from 23 stroke trials (25,674 patients) focusing on functional outcomes.
• Compared unadjusted and adjusted ordinal logistic regression models using simulated data.
• Assessed treatment effects using odds ratios (ORs) of 0.95, 0.74, and 0.57.

Main Results:

• Adjusting for baseline factors (age, sex, severity) consistently reduced sample size requirements across all assessed treatment effects.
• Median percentage reductions in sample size ranged from 35.3% to 38.4%.
• The proportion of simulations showing a greater treatment effect with adjusted models increased with stronger treatment effects.

Conclusions:

• Adjusting for prognostic factors in stroke trials offers substantial sample size reductions (at least 20-30%) for a given statistical power.
• This statistical adjustment can enhance the power of analyses, potentially aiding in the detection of effective interventions.
• Consider powering trials for unadjusted analyses and subsequently increasing power through prognostic factor adjustment.