Comparison of the power and type 1 error of total score models for drug effect detection in clinical trials
2025

Comparing Models for Detecting Drug Effects in Clinical Trials

Sample size: 12 publication 10 minutes Evidence: high

Author Information

Author(s): Haem Elham, Karlsson Mats O., Ueckert Sebastian

Primary Institution: Shiraz University of Medical Sciences, Shiraz, Iran; Uppsala University, Uppsala, Sweden

Hypothesis

How do different total score models compare in their ability to detect drug effects in clinical trials?

Conclusion

The I-BI model is the most effective for detecting drug effects when an IRT model is available, while the BI model is preferable when it is not.

Supporting Evidence

  • The I-BI model showed the highest statistical power and minimal bias.
  • Total score models were compared using simulated data from a clinical trial.
  • Different sample sizes were tested to evaluate model performance.
  • Mis-specified item parameters affected the power and type 1 error of IRT-informed models.

Takeaway

This study looked at different ways to analyze scores from clinical trials to see which method works best for finding out if a drug is effective.

Methodology

Data was simulated from a clinical trial setting using an IRT model, and the power, type 1 error, and treatment effect bias of various total score models were compared.

Potential Biases

Potential bias exists if the IRT model is not accurately specified.

Limitations

The study focused on simulated data from a specific IRT model in Parkinson's disease, which may not generalize to other diseases or models.

Participant Demographics

Participants were simulated in a clinical trial setting for Parkinson's disease.

Digital Object Identifier (DOI)

10.1007/s10928-024-09949-0

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication