Analysis of National Drug Code Identifiers in Ambulatory E-Prescribing.

BACKGROUND: Communication of an accurate and interpretable drug identifier between prescriber and pharmacist is critically important for realizing the potential benefits of electronic prescribing (e-prescribing) while minimizing its risk. The National Drug Code (NDC) is the most commonly used codified drug identifier in ambulatory care e-prescribing, but concerns have been raised regarding its use for this purpose. 

OBJECTIVES: To (a) assess the frequency of NDC identifier transmission in ambulatory e-prescribing; (b) characterize the type of NDC identifier transmitted (representative, repackaged, obsolete, private label, and unit dose); and (c) assess the level of agreement between drug descriptions corresponding to NDC identifiers in electronic prescriptions (e-prescriptions) and the free-text drug descriptions that were entered by prescribers. 

METHODS: We analyzed a sample of 49,997 e-prescriptions that were transmitted by ambulatory care prescribers to outlets of a national retail drugstore chain during a single day in April 2014. The First Databank MedKnowledge drug database was used as the primary reference data base to assess the frequency and types of NDC numbers in the e-prescription messages. The FDA's Comprehensive NDC Standard Product Labeling Data Elements File and the National Library of Medicine's RxNorm data file were used as secondary and tertiary references, respectively, to identify NDC numbers that could not be located in the primary reference file. Three experienced reviewers compared the free-text drug description that had been entered by the prescriber with the drug description corresponding to the NDC number from 1 of the 3 reference database files to identify discrepancies. Two licensed pharmacists with residency training and ambulatory care experience served as final adjudicators.

RESULTS: A total of 42,602 e-prescriptions contained a value in the NDC field, of which 42,335 (84.71%) were found in 1 of the 3 study reference databases and were thus considered to be valid NDC numbers. A total of 28,172 (67.70%) e-prescriptions in the sample were found to contain a representative NDC number, according to the definition used by the National Council for Prescription Drug Programs (NCPDP). The remaining e-prescriptions consisted of 4 subtypes of unrepresentative NDC numbers. In 41,298 (97.55%) e-prescriptions that contained an NDC number, the drug description associated with the number from 1 of the 3 data source files pointed to the identical semantic drug concept as the free-text drug description that had been entered by the prescriber. However, in 87 (0.21%) e-prescriptions, the free-text drug descriptions and the drug description associated with the NDC number pointed to completely different semantic drug concepts.

CONCLUSIONS: We found the use of NDC identifiers in our sample of e-prescriptions to be relatively high. However, approximately one-third consisted of unrepresentative NDC numbers (obsolete, repackaged, unit dose, or private label) that have the potential to create workflow disruptions at the dispensing pharmacy. Most disturbing was our finding that more than 2 out of every 1,000 e-prescriptions in our sample contained a free-text drug description that pointed to a completely different drug concept than that associated with its NDC value. Our study suggests the need for e-prescribing technology vendors to maintain accurate and up-to-date drug database files within their systems and to conduct regular validation checks to ensure that the drug descriptions associated with the NDC identifier and the free-text drug description that is sent in the e-prescription message point to the same drug concept. The FDA may need to consider a more active role in ensuring the accuracy of NDC assignment by drug manufacturers.