Better Communication, Better Care Using SICG Assistant Based on Artificial Intelligence (SICG-Assist)

Communication is a cornerstone of person-centered care, and takes on particular importance in serious illness, defined here as a condition carrying a high risk of mortality that negatively affects quality of life or daily functioning and that commonly produces significant emotional distress in caregivers. Inadequate, delayed, or infrequent communication contributes to avoidable patient and family suffering, and clinicians often lack structured tools and training to conduct high-quality conversations about goals, values, and preferences in serious illness.

The Serious Illness Conversation Guide (SICG), originally developed and validated internationally, provides a structured framework that clinicians can use throughout the course of an illness to elicit patient values and preferences and align care accordingly. This project - conducted by a research team that collaborates directly with the guide's original international developer group - will implement the SICG in routine clinical practice across primary care and hospital-based palliative care services in the province of Málaga, Spain, supported by a purpose-built mobile application and virtual assistant designed to facilitate documentation and continuity of the conversation across multiple involved clinicians.

The study combines four complementary components:

Cluster-randomized controlled trial (quantitative, confirmatory): Healthcare professionals (HCPs) are randomized (block randomization, blocks of 4) into an intervention arm (trained in and using the SICG, supported by the app/virtual assistant) or a control arm (usual care). Patients and families are recruited by their assigned HCP and constitute the unit of analysis (multilevel/mixed-model analysis accounting for clustering at the HCP level).

Before-after evaluation of HCP communication skills and satisfaction following SICG training.

Feasibility and usability study of the mobile application and virtual assistant, using the System Usability Scale (SUS, Spanish-validated) and the mHealth App Usability Questionnaire (MAUQ).

Cost-utility analysis from the funder's perspective (1-year time horizon), estimating incremental cost-effectiveness based on training costs, implementation costs, and healthcare resource utilization (emergency and hospital admissions, palliative care referrals).

A parallel qualitative strand (individual/group narrative interviews with HCPs, patients, and family key informants) will explore perceived utility of the SICG and document emergent experiences during implementation, analyzed thematically using an iterative, constructionist approach (NVivo).

Ethical approval for evaluation of the SICG implementation has been granted by the Comité de Ética de Investigación Provincial de Málaga (session of 30/09/2021); an amendment is planned to extend approval to cover use of the supporting mobile application (feasibility study component).

Study Design Field Value Study Type Interventional Study Design Pragmatic cluster-randomized controlled trial (mixed-methods) Allocation Randomized (block randomization at HCP level, blocks of 4) Intervention Model Parallel assignment Randomization Unit Healthcare professional (cluster) Unit of Analysis Patient Masking None (open-label; not feasible to blind HCPs to guide use) Primary Purpose Supportive care / Health services research Estimated Cluster Size ~5 patients per HCP Number of Clusters 40 HCPs (20 intervention / 20 control) Condition/Disease

Serious illness (advanced/life-limiting disease with high mortality risk and significant impact on quality of life); Palliative Care

Intervention

Experimental arm: Complex two-phase intervention - (1) Training of HCPs in use of the SICG (structured communication skills course); (2) Use of the SICG by trained HCPs with patients and their families (approx. 5 patients/families per HCP), supported by a dedicated mobile application and virtual assistant.

Control arm: Standard/usual care (no SICG training or use).

Eligibility Criteria

Patients:

Serious illness as defined above Age 18-80 years Spanish as native language Karnofsky Performance Status ≥50, or ECOG Performance Status ≤2, or Palliative Performance Scale (PPS) ≥50 Signed informed consent

Family members:

Family member of an eligible patient with serious illness Age 18-80 years Spanish as native language Signed informed consent

Exclusion criteria (patients and family members): Cognitive or perceptual impairment precluding completion of study questionnaires.

Healthcare professionals (HCPs):

Inclusion: Specialists in Palliative Care, Oncology, Hematology, or Primary Care, affiliated with Fundación CUDECA, Servicio Andaluz de Salud, or Institut Català d'Oncologia Exclusion: Any professional not meeting inclusion criteria Sample Size

190 patients with serious illness, calculated to detect a between-group difference of 2.5 points in anxiety/depression scores, with 95% confidence and 80% power, accounting for a design effect of 1.2 and an intracluster correlation coefficient (ICC) of 0.05. Patients recruited via 40 HCPs, average cluster size of 5.

Outcome Measures

Primary Outcome:

Change in anxiety and depression levels (Hospital Anxiety and Depression Scale, HADS) in patients and family members, intervention vs. control group.

Secondary Outcomes:

Perceived usefulness of the SICG (patients, families, HCPs) HCP communication skills and self-efficacy (validated communication self-confidence and expected-outcomes scales; psychosocial aspects of patient care scale) SICG implementation process fidelity and reach (patient identification/referral criteria; number of eligible vs. enrolled patients) Quality of life (EuroQoL-5D-5L; QLQ-C15-PAL) Cost-utility of SICG implementation (incremental cost-effectiveness ratio) Additional patient measures: Karnofsky Index, PPS, PS, Edmonton Symptom Assessment Scale (ESAS), Distress Thermometer App/virtual assistant usability and feasibility (SUS, MAUQ; learnability, efficiency, error rate, user satisfaction, engagement) Data Collection & Analysis

Quantitative data collected via electronic case report form (REDCap, hosted by Instituto IBIMA). Qualitative data collected via audio-recorded interviews, transcribed and analyzed in NVivo. Quantitative analysis by intention-to-treat; multilevel modeling to account for clustering at the HCP level; ICC calculated for the primary outcome; repeated-measures ANOVA for within-group change over time; multivariate imputation for missing data; significance threshold p<0.05 (two-sided); SPSS v25.0.

Estimated Timeline Recruitment of HCPs and training: approx. 8 months SICG implementation and follow-up: minimum 12 months per participating HCP Utility interviews: two time points during implementation Total estimated project duration: 36 months Data Sharing / Repository

De-identified data intended for deposit in a public repository (RIUMA - Universidad de Málaga institutional repository) following FAIR principles, subject to data protection regulations (GDPR; Spanish LOPDGDD 3/2018). Minimum 10-year data availability guaranteed by the research team.