Pseudo cluster randomization performed well when used in practice.

Area of Science:

• Clinical Trials Methodology
• Health Services Research

Background:

• Cluster randomization risks selection bias, while individual randomization risks contamination.
• Pseudo cluster randomization (PCR) was developed to mitigate these issues in clinical trials.

Purpose of the Study:

• To evaluate the performance of pseudo cluster randomization (PCR) in the Dutch EASYcare Study.
• To assess if PCR maintained group comparability and avoided unblinding or selection bias.

Main Methods:

• Clinicians were randomized using PCR into high (H) or low (L) intervention proportion groups.
• Data collected included clinician preferences, recruitment behavior, contamination, and perceived allocation ratios.
• Patient baseline characteristics and clinician recruitment rates were compared between groups.

Main Results:

• Groups were comparable at baseline.
• Clinicians showed a preference for the intervention arm.
• 58% of clinicians would recruit fewer patients in a control-only scenario, and 65% used intervention elements in control patients.
• 67% of clinicians incorrectly estimated a 50:50 allocation ratio.

Conclusions:

• The underlying assumptions of PCR were largely met in this study.
• PCR performed satisfactorily, demonstrating no signs of unblinding or selection bias.
• PCR is a viable alternative to traditional randomization methods in certain clinical settings.