Keywords: Systematic review; health literacy; cancer treatment and care; primary care; Artificial Intelligence.

Background:

Health literacy (HL) is crucial for cancer care outcomes, as limited HL correlates with poor adherence and increased disparities. However, evidence on interventions remains fragmented: comprehensive synthesis examining intervention effectiveness, assessment tools, and contextual factors across the cancer continuum is lacking. Understanding enablers and barriers is essential for developing sustainable, evidence-based strategies.

Research questions:

Across the cancer continuum:
• What is the effectiveness of HL interventions?
• Which validated tools measure HL domains?
• What barriers and enablers influence intervention effectiveness?

Method:

This PRISMA-compliant systematic review (PROSPERO: CRD420251165432), part of iNTERVENE (GA 101219203, Call EU4H 2024 PJ 02), searched six databases (2016-2025) for controlled trials examining HL interventions in adults across cancer continuum. From 600+ records, two independent reviewers screened titles/abstracts: 214 advanced to full-text assessment (212 retrievable). AI-assisted methodology employed Mistral Teams for full text data extraction and bias assessment (RoB-2 for RCTs; ROBINS-I V2 for single-arm studies), with human verification ensuring accuracy. Interventions classified using WHO-ICHI framework. Random-effects meta-analysis when homogeneity permits, otherwise narrative synthesis. GRADE assesses evidence certainty.

Results:

AI-assisted text extraction is ongoing using Mistral Teams (pilot validation: 12 papers with 28 variables of interest; accuracy=80%, precision=80%, recall=97%, F1=87%). Preliminary screening reveals heterogeneity across intervention modalities (educational programs, professional training, peer support…), target populations (patients, caregivers, healthcare professionals…), validated assessment instruments, cancer types, and care continuum phases (prevention, diagnosis, treatment…). Intervention intensity varies from single-session to comprehensive year-long programs. Narrative synthesis will characterize all included studies; meta-analyses or statistical pooling will be conducted if sufficient studies with similar HL outcomes are available. Complete results are expected by end of April 2026.

Conclusions:

This systematic review will provide comprehensive evidence on HL intervention effectiveness across cancer continuum, identifying optimal implementation strategies to reduce disparities and improve outcomes in primary care oncology across diverse European healthcare contexts.

Points for discussion:

How can primary care physicians integrate validated HL assessment tools into routine cancer care consultations without increasing consultation time burden?

How can findings from this systematic review be translated into practical, sustainable interventions implementable within resource-constrained primary care systems across different European healthcare contexts?

Can AI models nowadays be trusted as independent reviewers in systematic reviews, replacing traditional dual human screening and extraction processes? What quality benchmarks should guide this already ongoing transition?