Abstract
BACKGROUND: Patient portal secure messaging allows patients to describe health-related behaviors in ways that may not be sufficiently captured in standard electronic health record (EHR) documentation, but little is known about how cannabis is discussed on this platform. OBJECTIVE: This study aimed to identify patient and provider secure messages that discussed cannabis and contextualize these discussions over periods before and after its legalization for medical purposes in Pennsylvania. METHODS: We examined 382,982 secure messages sent by 15,340 patients and 6101 providers from an integrated health delivery system in Pennsylvania, United States, from January 2012 to June 2022. We used an unsupervised natural language processing approach to construct a lexicon that identified messages explicitly discussing cannabis. We then conducted a qualitative content analysis on a random sample of identified messages to understand the medical reasons behind patients' use, the primary purposes of the cannabis-related discussions, and changes in these purposes over time. RESULTS: We identified 1782 messages sent by 1098 patients (7.2% of total patients in the study) and 800 messages sent by 430 providers (7% of total providers in the study) as explicitly discussing cannabis. The most common medical reasons for use stated by patients in 190 sampled messages included pain or a pain-related condition (50.5% of messages), anxiety (13.7% of messages), and sleep (11.1% of messages). We coded 56 different purposes behind the mentions of cannabis in patient messages and 33 purposes in 100 sampled provider messages. In years before the legalization (2012-2016), patient and provider messages (n=20 for both) were primarily driven by discussions about cannabis screening results (38.9% and 76.5% of messages, respectively). In the years following legalization (2017-2022), patient messages (n=170) primarily involved seeking assistance to facilitate medical use (35.2% of messages) and reporting current use (25.3% of messages). Provider messages (n=80) were driven by giving assistance with medical marijuana access (27.5% of messages) and stating that they were unable to refer, prescribe or recommend medical marijuana (26.3% of messages). CONCLUSIONS: Patients showed a willingness to discuss cannabis use over patient portal secure messages and expressed interest in use after the legalization of medical marijuana. Some providers responded to patient inquiries with assistance in obtaining access to medical marijuana, while others cautioned patients on the risks of use. Insight into cannabis-related discussions through secure messages can help health systems determine opportunities to improve care processes around patients' cannabis use, and providers should be supported to communicate accurate and consistent information.