Connectional Silence in Telemedicine-Facilitated Palliative Care Conversations (GP113)

Outcomes 1. Participants will be able to define the concept of connectional silence (CS) and identify the criteria used to subcategorize CS in the context of telemedicine palliative care consultations.2. Participants will understand the differences in telemedicine versus in person connectional silence (CS) and the impact on measurement of CS with human coding and computer algorithms. Key Message Despite its significance, research on connectional silence (CS) in telepalliative care is limited. This study addresses this gap. Combining human coding and machine learning, it explores CS prevalence and subtypes in telehealth, offering insights into enhancing empathetic care delivery. Importance Connectional silence (CS) in palliative care conversations fosters comfort and trust and is linked to better quality of life and decision-making. COVID-19 highlighted the role of telehealth however it is unclear if CS is present or different in telehealth compared to in-person communication. Objective(s) We sought to determine the prevalence of CS in telepalliative care using dual human and computer approaches. Scientific Methods Utilized We studied participants from the “Telemedicine facilitated palliative care consultations in rural dialysis study,” who were recruited between 2018-20 from five dialysis units. A human coder identified silence defined as pauses of ≥ 2 seconds with no utterances from audio files. Pauses were double-coded as CS and sub-categorized as invitational, emotional, or compassionate. Next, CS were identified using an established computer algorithm based on convolutional neural networks and random forest machine learning methods. Results 39 participants (90% non-Hispanic White, 56% male) were recruited and 34 participants completed a telepalliative care consultation recorded via Zoom. Intra-rater reliability for CS was 93.5%. Human coding identified 666 pauses of ≥ 2 seconds and computer coding identified 360 pauses of ≥ 2 seconds, with 151 pauses identified by both methods. Among human-identified pauses, 57 (8.6%) were CS. There were 43 (75.4%) invitational CS and 10 (17.5%) emotional CS. Conclusion(s) CS was identified using human and established computer algorithms during telehealth palliative care consultations with patients receiving dialysis. The prevalence of CS in these teleconsults was similar to a reported prevalence for in-person palliative care. Invitational CS was more common in televideos than in prior work, which may be due to the conversational and technical aspects of telehealth. Impact We confirm the presence of CS in telepalliative care conversations using human and computer coding approaches, which supports the use of telehealth for palliative care delivery in rural settings.