Table 5.

False negative errors made by the medication discrepancy detection algorithm

Cause of false negative errors identified by the chart review	Error [%]
1. The medication was omitted by the medication entity detection algorithm	68.0%
2. The medication was matched to a wrong medication due to similar medication names (e.g. methylprednisolone and prednisolone)	9.1%
3. The prescription contains more ingredients than the medication in the clinical note or vice versa (e.g. albuterol vs. ipratropium albuterol)	6.3%
4. The medication in the clinical note was matched to a correct prescription (e.g. matching diastat to diazepam) but the prescription had a different route (e.g. oral route vs. rectal route)	6.0%
5. The medication and the prescription names co-occurred in the same RxNorm description as ingredients rather than synonyms (e.g. “glycerin” and “polyethylene glycol” co-occurred in the RxNorm description of “artificial tears”)	4.8%
6. Other reasons	5.8%