JMIR Publications

ABSTRACT

Background:

Telehealth interventions where providers offer support and coaching for patients with chronic conditions such as heart failure (HF) and diabetes (T2DM) are effective in improving health outcomes. However, the question of how the content and structure of these interactions relate to healthcare utilization is not well understood.

Objective:

We aimed to investigate how the content and structure of telesupport conversations relate to hospital utilization in patients with cardiometabolic disease. Specifically, we focused on conversations pertaining to lifestyle and medication management (LM) topics.

Methods:

We leveraged real-world data from 50 HF patients enrolled in a post-discharge telesupport program with primary intervention comprising a series of telephone calls from nurse telecarers over a 12-month period. For the full cohort, we transcribed 729 English calls and annotated conversation topics. For a sub-cohort (25 patients) with both HF and T2DM, we also annotated LM-related content with dialogue acts describing typical conversational structures. For each patient, we identified calls with unusually high ratios of utterances on LM topics as LM-focused calls. We further extracted utilizations data for inpatient admissions between 6 months before to 6 months after the intervention period. Firstly, we assessed associations between the number of LM-focused calls and the number of admissions and average length of stay (LOS) per admission. Secondly, to understand telecarer perspectives on these calls, we conducted an expert evaluation where two telecarers judged levels of concern and follow-up actions for LM-focused and non LM-focused calls. Finally, to understand conversational structures and content of LM-focused calls, we compared the number of speaker utterances, dialogue acts, and symptom attributes in LM-focused calls to those in calls containing, but not focused, on LM content.

Results:

The number of lifestyle-focused calls a patient had was associated with lower average LOS for inpatient admissions (Spearman ρ=-0.27, P=.03), but not with number of admissions. There were no significant associations linking other calls to admissions. In the expert evaluation, telecarers identified 24.2% of calls assessed (29/120) as concerning enough for follow-up. For these 29 calls, telecarers were more attuned to concerns on symptoms and vitals (19/29 calls) than to lifestyle and medication management concerns (7/29 calls). In conversational analyses, LM-focused calls had significantly fewer utterances (P<.001, Cohen d=-0.56) and more dialogue acts (P<.001, Cohen d=0.76) than calls containing but not focused on LM content. LM-focused calls did not contain deeper discussions on clinical symptoms. These findings indicate that LM-focused calls entail short, intense discussions with greater emphasis on understanding patient experience and coaching than on clinical content.

Conclusions:

Our work offers novel characterization of telesupport intervention strategies and their association with healthcare utilization among patients with cardiometabolic disease. Our approach and findings could inform ways to enhance telehealth programme content for better patient outcomes.