A Quantitative Analysis of Patient Pronoun Use in Medical Narratives

Abstract

Analysis of patient language within illness narratives has been helpful to understand how patients view their illnesses and differences in how patients and doctors understand specific illness processes. While substantial work has been done in this area, particularly in the area of metaphor analysis, less research has analyzed disease referents and how they change over the course of an unstructured patient narrative in order to investigate how a patient’s relationship to their illness changes depending on context. We collected and transcribed interviews from a small cohort of six individuals with histories of an illness process not restricted to a single disease category. Through discussion between two readers, we identified and categorized all medical disease referents and associated words (possessives and demonstratives) in the interview transcripts into medical disease referents and medical abbreviations, pronouns and other and into different contexts from within the narrative (pre- vs. post diagnosis, medical vs. non-medical, moments of metaperspective-taking and moments of empowering self-dialogue), after which we performed goodness of fit statistics to determine which categories and associated words occurred more frequently in which contexts. We found that compared to all other disease referents, medical disease referents and medical abbreviations were significantly more likely to occur in medical contexts and were significantly less likely to be attached to the words “this” or “these,” words defined as “indicating a thing or person present or near (actually in space or time, or ideally in thought)” according to the Oxford English Dictionary (OED). Conversely, it was found that contexts involving moments of meta-perspective- taking on illness as well as moments of empowering self-dialogue had significantly more use of “this” or “these” acting as a disease referent on its own and attached to disease referents than in other contexts. Overall, the results point to restriction of medical terms for illness within medical contexts and decreased use of words indicating closeness to the speaker for medical terms. Implications of this early study include a novel framework for future analysis of unstructured patient narratives.