Patient experience literature in early-stage breast cancer (eBC) is limited. This study used a mixed-methods approach to examine patient conversations from public online forums to identify and evaluate eBC-related themes. Among 60,000 eBC-related posts published September 2014-2019, text from a random subset of 15,000 posts was extracted and grouped into linguistically similar, mutually exclusive clusters using an advanced natural language processing (NLP) algorithm. Clusters were characterized using four quantitative metrics: betweenness centrality (linguistic similarity to other areas of the cluster network), sentiment (general attitude toward a topic), recency (average date of posts), and volume (total number of posts). This analysis represented 3906 unique users (67% and 33% obtained from cancer-specific and general health/nonhealth forums, respectively). Of the 27 clusters identified, most important were "discussing recurrence & progression," "understanding diagnosis & prognosis," and "understanding cancer, biomarkers, and treatments." Several major themes related to recurrence risk, diagnosis, monitoring, and treatment were identified. Additional emphasis on communicating the disease recurrence risk and shared decision-making could strengthen patient-clinician partnerships.
