Implementing Electronic Health Record Default Settings to Reduce Opioid Overprescribing: A Pilot Study.

Objective: To pilot test the effectiveness, feasibility, and acceptability of instituting a 15-pill quantity default in the electronic health record for new Schedule II opioid prescriptions. Design: A mixed-methods pilot study in two health systems, including pre-post analysis of prescribed opioid quantity and focus groups or interviews with prescribers and health system administrators. Methods: We implemented a 15-pill electronic health record default for new Schedule II opioids and assessed opioid quantity before and after implementation using electronic health record data on 6,390 opioid prescriptions from 448 prescribers. We then analyzed themes from focus groups and interviews with four staff members and six prescribers. Results: The proportion of opioid prescriptions for 15 pills increased at both sites after adding an electronic health record default, with one reaching statistical significance (from 4.1% to 7.2% at CHC, P = 0.280, and 15.9% to 37.2% at WVU, P < 0.001). The proportion of 15-pill prescriptions increased among high-prescribing departments and among most high- and low-frequency prescribers, except for low-frequency prescribers at CHC. Sites reported limited challenges in instituting the default, although ease of implementation varied by electronic health record vendor. Most prescribers were not aware of the default change and stated that they made prescribing decisions based on patient clinical characteristics rather than defaults. Conclusions: This pilot provides initial evidence that changing default settings can increase the number of prescriptions at the default level. This low-cost and relatively simple intervention could have an impact on opioid overprescribing. However, default settings should be selected carefully to avoid unintended consequences.

Assessment of commercial NLP engines for medication information extraction from dictated clinical notes.

PURPOSE: We assessed the current state of commercial natural language processing (NLP) engines for their ability to extract medication information from textual clinical documents. METHODS: Two thousand de-identified discharge summaries and family practice notes were submitted to four commercial NLP engines with the request to extract all medication information. The four sets of returned results were combined to create a comparison standard which was validated against a manual, physician-derived gold standard created from a subset of 100 reports. Once validated, the individual vendor results for medication names, strengths, route, and frequency were compared against this automated standard with precision, recall, and F measures calculated. RESULTS: Compared with the manual, physician-derived gold standard, the automated standard was successful at accurately capturing medication names (F measure=93.2%), but performed less well with strength (85.3%) and route (80.3%), and relatively poorly with dosing frequency (48.3%). Moderate variability was seen in the strengths of the four vendors. The vendors performed better with the structured discharge summaries than with the clinic notes in an analysis comparing the two document types. CONCLUSION: Although automated extraction may serve as the foundation for a manual review process, it is not ready to automate medication lists without human intervention.