Analysis of a State Controlled Substance Prescription Registry: Lessons Learned

Recent reports have detailed the increase in fatal prescription drug poisonings. We studied the relationship between apparently legitimate prescribing captured in the state controlled substance registry and fatal drug poisonings. The Utah Controlled Substances Database contains information about all filled prescriptions for scheduled drugs in non-inpatient settings. Clinicians use the data, and the Division of Occupational and Professional Licensing uses it to identify potential cases of drug over-utilization, misuse, and over-prescribing. Our study at the Utah Department of Health was the first analysis of the database for research purposes. We report lessons learned by adapting a database intended for enforcement to research. Pharmacies submit data electronically, and minimal quality checks occur (for missing data) before records are included. This level of scrutiny has been identified as appropriate for their enforcement mandate, but stricter evaluation would improve research utility. Non-missing but invalid values in key fields such as patient name, drug code, or provider DEA number compromise individual-level analysis. Each prescription is a separate record in the database. Creation of a master patient index would facilitate generation of longitudinal patient histories — possible now only with probabilistic linking techniques. The current database accepts direct exports from the pharmacies’ usual data collection system, so some features that would be useful for research were not feasible when the database was created. Jurisdictions creating a controlled substance prescription registry de novo should consider using legal values for data entry fields, including indicator variables for non-human animal prescriptions and prescriptions picked up by proxy, and standardization of acceptable values within name fields. Research using prescription registries may help improve their surveillance and enforcement functions by identifying risk factors for adverse events and indicators of possible fraud or inappropriate prescribing. This project was part of the Utah Center of Excellence in Public Health Informatics.