Data Mining FDA Docket 2019-N-1482: Content, Sentiment, and Metadata

Abstract

The legal status of cannabis continues to evolve, raising challenges for medical writers who work in population health and drug safety. To guide messaging, research has investigated how the public perceives cannabis, often relying on surveys or “big data” analyses of social media. However, these methods can be costly. As a supplement, we explored comments posted to a United States Food and Drug Administration docket on cannabis science and risk, which may offer an accessible, purposive, cost-effective source of data. We applied a multipronged methodology that involved content analysis, sentiment analysis, and metadata analysis. The findings suggest that broad messaging on cannabis may have limited effectiveness. Instead, medical writers should design messages that emphasize the risks of particular products as well as express empathy for consumers suffering from specific conditions. Moreover, among other things, the findings suggest that medical writers should use the terms “cannabis” and “marijuana” intentionally, considering the implications of each. In the future, research should develop methods to further segment drug consumers demographically and psychographically, building on the methodology that we present here. This research may inform not just messaging but regulatory writing practices and state drug policies.