Detection and prediction limits for identifying highly confusable drug names from experimental data

Area of Science:

• Pharmacovigilance
• Cognitive Psychology
• Biostatistics

Background:

• Drug name confusion is a significant issue, leading to errors, costs, and patient harm.
• Current preapproval screening methods lack robust experimental designs and statistical validation.
• Reliable methods are needed to assess the confusability of proposed new drug names.

Purpose of the Study:

• To develop and validate experimental designs and statistical methods for preapproval screening of drug names.
• To establish a reliable threshold for identifying unacceptably confusing drug names.
• To improve the safety of new drug introductions through rigorous name evaluation.

Main Methods:

• Conducted four memory and perception experiments with clinicians using test and control drug names.
• Utilized logistic regression, Poisson prediction limits, and credible intervals for data analysis.
• Compared error rates of new drug names against a distribution of established drug names.

Main Results:

• Developed models with an excellent fit to experimental data.
• Demonstrated a method to identify a threshold for unacceptably confusing names.
• Established a framework for valid inferences on drug name confusability.

Conclusions:

• The described experimental designs and analytic methods provide a valid approach for preapproval drug name testing.
• These methods can be applied in regulatory settings to assess comparative safety and effectiveness.
• Implementation of these methods can reduce drug name confusion and improve patient safety.