Clinician Self-Referentiality

Background Language in clinical notes can serve as a proxy for the note writer's cognitive processes regarding a patient encounter. Use of personal pronouns (eg, I or we ) may represent agency or alliance in documentation. Research Question What factors are associated with personal pronoun use in ICU notes? Study Design and Methods We performed a retrospective cohort study of notes from adults (≥ 18 years of age) admitted to ICUs at the University of California, San Francisco (2012-2022). The primary outcome was personal pronoun frequency per note, modeled as a rate per sentence. Covariates included patient sociodemographics, author characteristics, gender concordance, and selected clinical characteristics. We performed a cluster-robust Poisson regression model, a sensitivity analysis limited to authors with prior personal pronoun use, and a subset analysis removing attending notes. Results Across 842,953 notes for 5,799 patients, personal pronouns were present in 53% of notes. Patient demographics and clinical factors were not associated meaningfully with personal pronoun use. Women authors were less likely than men authors to include personal pronouns. Residents used fewer personal pronouns than attendings. Notes from advanced practice providers, registered nurses, respiratory therapists, physical therapists, and occupational therapists were associated with fewer personal pronouns than those from doctors of medicine or doctors of osteopathic medicine. Pronoun use also varied by ICU type and note author specialty. Gender concordance was not associated with personal pronoun use. Select covariates varied by pronoun type (I vs we). Results were consistent in the sensitivity analysis, although they varied in the subset analysis. Interpretation Our results show that personal pronoun use in ICU notes is associated primarily with author characteristics. Future studies should explore author intent and reader interpretation.