Improving pragmatic clinical trial design using real-world data.

Area of Science:

• Clinical Trials
• Health Informatics
• Public Health

Background:

• Pragmatic clinical trials increasingly utilize automated data sources like electronic health records (EHRs) for participant identification and outcome collection.
• Automated data collection provides valuable insights during the trial design phase.
• This study outlines methods for effectively using historical data to inform trial design decisions.

Purpose of the Study:

• To demonstrate how historical EHR data can inform the design of pragmatic clinical trials.
• To illustrate the application of these methods in a suicide-prevention trial.
• To enhance the realism of power and sample size calculations.

Main Methods:

• Utilized historical EHR data from 122,873 individuals with Patient Health Questionnaire (PHQ) responses.
• Analyzed suicide attempt rates in the 18 months post-PHQ completion, focusing on PHQ item nine responses.
• Conducted simulations using EHR data to estimate statistical power for detecting a 25% reduction in suicide attempts.

Main Results:

• Suicide attempt rates varied based on PHQ item nine responses.
• Estimated a decrease in individuals with elevated PHQ scores (history of suicidal ideation) from 50% to 5% over 50 weeks.
• Simulation indicated 90% power to detect a 25% suicide attempt reduction with 8000 participants per arm (alpha=0.05).

Conclusions:

• Historical data is crucial for informing pragmatic clinical trial design, especially those using automated data collection.
• Real-world data enables more accurate sample size calculations and power estimations.
• Data-informed trial design optimizes recruitment efficiency and statistical power.