The extent to which medication prescriptions had a reason for the medication use documented in form of a potential indication within the electronic health record (EHR) problem lists using a MEDication Indication (MEDI) resource was evaluated.Methods
Prescriptions from January 1 to June 30, 2015, comparing them to patients' problem lists using MEDI and the MEDI High Precision Subset (MEDI-HPS) were analyzed. RxNorm generic ingredient name codes in MEDI were used to map prescriptions to problems using codes from the International Classification of Diseases, 9th edition. A reference standard was established to evaluate the MEDI precision and recall by having 2 pharmacists independently manually review prescriptions and problem lists from 30 randomly selected patients.Results
For 62,191 patients, 61% of prescriptions matched a potential indication on the patient's problem list using MEDI, whereas only 38% had a match using MEDI-HPS. The precision of MEDI compared to the reference standard was 47% with a recall of 57%, whereas MEDI-HPS had a precision of 79% and recall of 96%. Secondary analysis excluding medication prescribed with a supply of ≤14 days gave slightly better, yet not significant, results.Conclusion
Analysis of patient records found most patients did not have an indication listed in the EHR problem list that would match a specific medication, suggesting that the problem lists may be incomplete. When using MEDI, 61% of prescriptions matched to the problem list, compared with only 38% using MEDI-HPS. Likewise, 37% of problems matched to prescriptions when using MEDI, compared with only 21% using MEDI-HPS.