Identifying #addiction concerns on twitter during the COVID-19 pandemic: A text mining analysis

The 2019 Novel Coronavirus (COVID-19) is responsible for thousands of deaths and hospitalizations. To curb the spread of this highly transmissible disease, governments enacted protective guidelines for its citizens, including social distancing and stay-at-home orders. These restrictions on social interactions can be especially problematic for individuals managing or recovering from addiction given that treatment often involves access to services and resources that became limited or even unavailable at this time. Social media sites like Twitter serve as a space for users to post questions and concerns about timely topics and allow for researchers to track common themes among the public. The goal of this study was to identify how the public was discussing addiction on Twitter during the COVID pandemic. Methods: We performed a text mining analysis to analyze tweets that contained "addiction" and "covid" to capture posts from the public that illustrated comments and concerns about addiction during the COVID-19 pandemic. We report on 3,301 tweets captured between January 31 and April 23, 2020. The study was conducted in the United States, but contained tweets from multiple countries. Results: The most prevalent topics had to do with services offered by Acadia Healthcare and Serenity Healthcare Centers, attempts to manage time while home, difficulties of coping with alcoholism amidst rising sales of alcohol, and attention to ongoing health crises (e.g.,., opioids, vaping). Additional topics included affordable telehealth services, research from France on the relationship between nicotine and COVID-19, concerns about gambling addiction, and changing patterns in substance misuse as drug availability varies. Conclusions: Analyzing Twitter content enables health professionals to identify the public's concerns about addiction during the COVID-19 pandemic. Findings from text mining studies addressing timely health topics can serve as preliminary analyses for building more comprehensive models, which can then be used to generate recommendations for the larger public and inform policy.