Publication date

Exploring the relationship between GP and patient language use: the case of (un)explained symptoms.

Stortenbeker, I., olde Hartman, T., Stommel, W., Dulmen, S. van, Das, E. Exploring the relationship between GP and patient language use: the case of (un)explained symptoms. Patient Education and Counseling: 2023, 109(Suppl. S), p. 42-43. Meeting abstract

In previous research, we found that general practitioners (GPs) systematically vary their language use when patients present with medically unexplained symptoms (MUS) versus medically explained symptoms (MES). Yet, we did not observe such difference in various linguistic markers of patients with MUS versus MES. This raises the question whether and how GP and patient language use are related, e.g. are GPs’ diminishers (“this is a little red”) related to patients’ use of intensifiers (“it’s really painful”)? The current study therefore explores the overall relationship between GP and patient linguistic markers.

Observational study combining two annotated datasets of linguistic markers in GP and patient utterances in 82 consultations (41 MUS, 41 MES). We calculated relative frequency scores per consultation of GP linguistic markers (negations ["not"], intensifiers [“really”], diminishers [“a little”], uncertainty markers [“maybe”]). These scores were included as predictor variables in linear regression models with relative frequencies of various relevant patient markers (negations, intensifiers, diminishers, subjectivity markers [“I believe], 1st person subject [“me” or “my body”]) as outcome variables. We then added the interaction terms between GP markers and patients’ MUS or MES, and potentially relevant confounders in a step-wise approach.

We identified 2590 relevant GP utterances and 2752 relevant patient utterances. We observed no main effect of GPs’ language use on patients’ linguistic markers, and this appeared to be similar for patients with MUS or MES.

No overall systematic relationship between GP and patient language can be observed in a selection of linguistic markers. This means that linguistic markers used by patients does not depend on those used by GPs, e.g. GPs’ diminishers (“somewhat painful”) do not relate to patients’ intensifiers (“really awful”) at the consultation level. In this presentation, we will discuss the implications of combining two previously annotated datasets to generate new research questions.