Crossover and self-controlled designs in clinical research.

Area of Science:

• Clinical Trials Methodology
• Biostatistics

Background:

• Crossover and self-controlled study designs efficiently yield valid clinical trial results with reduced patient numbers.
• These designs are valuable for clinical research, but their effective implementation requires careful consideration of methodological aspects.

Purpose of the Study:

• To evaluate the design and analysis of crossover studies published in a specific journal between 1978-1979.
• To identify common practices and potential limitations in the application of crossover and self-controlled study designs.

Main Methods:

• Review of 13 crossover studies from 1978-1979, assessing key design features.
• Examination of patient assignment, treatment switching criteria, blinding, order effects, and statistical analysis.
• Brief review of 28 self-controlled studies.

Main Results:

• Only 7 of 13 crossover studies used random assignment for initial treatment.
• Most studies used time-dependent rules for treatment switching (10/13), but 3 used less appropriate disease-state-dependent rules.
• Blinding of the crossover point was infrequent (3/13), and assessment of treatment order effects was rare (1/13).
• Most studies (11/13) employed minimally acceptable statistical analysis.

Conclusions:

• Crossover and self-controlled studies show potential for efficient clinical research but require rigorous design and analysis.
• Methodological shortcomings, including randomization, blinding, and order effect assessment, were prevalent in the reviewed studies.
• Improvements in study design and statistical analysis are crucial for maximizing the validity and reliability of self-controlled and crossover trial findings.