Teaching Evidence Based Practice in 2026: it's time to rethink what and how we ask health professionals to learn

Evidence-based practice (EBP) education has long been anchored to a model designed for a different era, one in which practitioners were expected to routinely search for, retrieve and critically appraise individual primary studies. Whilst that knowledge and skills still matter, it no longer reflects how many health professionals actually access and use evidence in their practice in 2026.

In modern practice, health professionals seeking to inform their decisions with evidence rarely begin with a primary research study. In reality, they will most likely consult a clinical practice guideline, a point-of-care evidence summary, or, with growing frequency, outputs generated by large language model-based tools. This shift does not represent a failure of clinical culture; in most situations, it reflects appropriate and efficient use of the evidence ecosystem as it currently exists.

EBP education has been slow to adapt. Our paper, published in BMJ Evidence-Based Medicine (Innocenti et al., 2026, doi:10.1136/bmjebm-2025-114184), proposes a pragmatic recalibration of EBP education priorities. This is not a wholesale reinvention of the field, but a sharper focus on a small set of high-impact, observable competencies that reflect modern evidence use in decision making.

A focused competency set for the current evidence ecosystem
Rather than maintaining expansive lists of discrete EBP skills/competencies, we suggest that a modern baseline curriculum should ensure all learners can:

• Identify relevant preappraised sources: this includes guidelines, systematic reviews, point-of-care summaries that align for the health care question at hand.
• Judge their trustworthiness: the ability to assess whether a source is current and based on a well-conducted evidence syntheses, whether there is a clear presentation of benefits and harms, with appropriate certainty ratings, and, where applicable, a justified rationale for the strength of any recommendation(s).
• Interpret effect estimates and certainty assessments: for patient-important outcomes, with emphasis on absolute rather than relative effects, and recognising what the certainty of evidence will, and won’t, allow us to conclude.
• Recognise preference-sensitive decisions: and apply this understanding to support shared decision making.
• Verify AI-mediated outputs: by tracing key claims to trustworthy, current sources, before using outputs in practice.

This last competency warrants specific attention. Generative AI tools are already embedded in the environments where EBP is learned and practised. They can support clinical learning in meaningful ways including; simulating scenarios, extracting structured information, summarising complex evidence. However, there use can also introduce risks that educators must address directly including; unclear provenance, outdated or unsupported claims, and outputs that project authority while omitting uncertainty or harms. Importantly, AI verification should not be treated as a standalone advanced skill; it is better understood as an extension of the trustworthiness checks that effective EBP teaching should already prioritise. This does not make critical appraisal obsolete; rather, it makes it tiered: all learners need efficient trustworthiness checks, while deeper appraisal skills remain essential when trustworthy guidance is unavailable, outdated, inconsistent or not applicable.

Teaching and assessment implications
Defining what to teach is insufficient without addressing how to teach it and how to evaluate whether it has been learned. We advocate for decision-centred learning embedded in practice workflow, where educators make their reasoning visible; narrating their choice of source, their assessment of trustworthiness, their interpretation of effects, and the role of patient preferences in the decision. Evidence interpretation should be treated as an observable practice behaviour, practised repeatedly in realistic scenarios. A need to share such examples of this teaching model should now be the focus of the EBP educator community.

The notion of assessment for, as and of learning can be utilised as an interplay to a more holistic learning environment, one where authentic assessment is the centrepiece. Curricula evaluated primarily through knowledge-based tests and examinations will continue to teach to the test. A 2026-oriented assessment approach should instead examine whether learners can demonstrate the behaviours that matter in practice: judging trustworthiness, interpreting effects and certainty, recognising when shared decision making is required, and appropriately verifying AI-generated information before applying it to care.

A milestone at the EBHC Conference 2026
This paper is intended as a starting point for a broader conversation about the Sicily Statement update. Translating these priorities into updated, validated competency frameworks — with agreed definitions, teaching strategies, and assessment tools — represents the substantive work ahead.
The EBHC International Conference 2026 will provide a natural forum to take this discussion forward. David Nunan (Centre for Evidence-Based Medicine, University of Oxford) will deliver a dedicated keynote addressing the challenge of EBP education in the current evidence ecosystem — examining how competency frameworks, teaching strategies, and programme evaluation need to evolve in response to preappraised evidence, variable trustworthiness, and AI-mediated access to information.
The conference represents an important opportunity for educators, researchers and clinicians to engage with this agenda collectively.

We look forward to continuing this discussion in Sicily.