Validation and enhancement of a computable medication indication resource (MEDI) using a large practice-based dataset

Area of Science:

• Clinical Informatics
• Pharmacovigilance
• Health Services Research

Background:

• Accurate medication-indication linkage is crucial for clinical practice and research.
• Existing resources lack comprehensive, computable medication-indication data.
• The MEDI (Medication-Indication) resource was developed to address this gap.

Purpose of the Study:

• To evaluate the medication coverage of the MEDI resource using real-world clinical data.
• To calculate the prevalence of indications for each medication within a large patient cohort.
• To assess the clinical utility of MEDI for informatics and phenotype recognition.

Main Methods:

• Extracted medication and diagnosis data for 1.3 million individuals from Vanderbilt University Medical Center.
• Assessed MEDI's coverage of recorded medications and mapped indications to ICD-9 codes.
• Conducted manual physician review for four example medications to validate indications and prevalence estimates.

Main Results:

• MEDI demonstrated high medication coverage, linking 97.3% of medications found in electronic health records.
• A 'high precision subset' of MEDI covered 93.8% of recorded medications.
• Manual review confirmed MEDI's accuracy in identifying indications and estimating medication prevalence.

Conclusions:

• MEDI provides comprehensive and accurate medication-indication linkage, essential for clinical informatics.
• The inclusion of indication prevalence data enhances MEDI's clinical usability.
• MEDI is a valuable tool for electronic health record-based research, aiding in phenotype recognition.