Abstract
Background Prescription drugs are identified in the VDW pharmacy file by National Drug Code (NDC.) Although the NDC constitutes a standard nomenclature, there is a no single, comprehensive resource for codes, either in the HMORN or beyond. Instead, each site maintains an “everNDC” file based on its own internal and external sources. Therefore, to identify NDC codes for drugs of interest for a multi-site study, the recommended process (VDW EverNDC Data Structure, Version 3) involves the collection and compilation of codes from each site. This is a frequent and time-consuming practice. We will present an exploration of the benefits of this approach in the experience of the Mental Health Research Network (MHRN.)
Methods We will compare the results of the recommended practice to those from a shortcut limited to RXnorm (RXN), an open-source library from the National Library of Medicine and First DataBank (FDB), a proprietary database licensed by Group Health, for antidepressants, lithium, anticonvulsant mood stabilizers, first and second generation antipsychotics, benzodiazepines, and stimulants.
Results In our process of identifying antidepressant NDCs from a list of 32 active ingredient names, RXN and FDB generated a list of 15401 codes. At Group Health (GH), our historical NDC file had 1653 additional codes. However, these additional codes only match 35 fills in our pharmacy file, the most recent being in 2002. For comparison, the full list of antidepressant NDC codes yielded an average of 480,000 fills per year at GH over the period 2000–2009. From nine other sites, we received an additional 1329 codes, increasing the number of codes by about 8%.
Conclusions Our initial results indicate that for antidepressants, site-specific NDC codes at GH represent rare and older prescriptions. The RXN/FDB method yielded ~90% of the total NDC codes identified at our site, and >99.99 % of the prescriptions identified. We need to confirm this across other sites and other drug classes, but it is possible that a simplified process for defining a list of NDC codes for a drug class would be suitable for many multi-site studies.
