European clinicians have long relied on a small number of trusted clinical decision support tools to answer point-of-care questions quickly and safely. For many, OpenEvidence became part of that toolkit — a fast, AI-powered clinical search platform grounded in peer-reviewed literature. But in early 2026, OpenEvidence withdrew its app from the UK and European markets, a withdrawal widely attributed to regulatory uncertainty surrounding the EU Artificial Intelligence Act and its implications for AI tools that influence clinical decision-making. That withdrawal has left a practical gap for European clinicians and raised a wider question: what does a genuinely Europe-appropriate clinical decision support tool actually look like?

Why European clinicians are reconsidering their options

The timing of OpenEvidence's exit is not coincidental. The EU AI Act classifies AI-based clinical decision support systems as high risk, triggering a set of transparency, provenance, and robustness obligations that go well beyond what most US-built tools were designed to meet. Alongside this, the General Data Protection Regulation (GDPR) continues to impose strict requirements on where patient data is processed and stored, and the EU Medical Device Regulation (MDR) sets a distinct regulatory bar for software that influences clinical decisions.

Nearly three quarters of EU countries are already using AI-assisted diagnostics, including tools that support clinical decision-making, so the demand for these technologies is not in question. What is in question is whether the tools clinicians reach for are built to operate safely and lawfully within European healthcare systems.

OpenEvidence is a clinical decision support platform used in the US market across multiple hospitals. Its absence in Europe creates a moment for European clinicians to evaluate alternatives with clearer eyes.

What to look for in a clinical decision support tool as a European clinician

Before evaluating any specific platform, it helps to have a clear framework for what matters in the European context. The criteria that most commonly distinguish appropriate tools from inappropriate ones for European use are:

Regulatory compliance

• Is the tool registered or certified under EU MDR or the equivalent national framework (for example, MHRA registration in the UK)?

• Has it been assessed under the EU AI Act's high-risk AI provisions, including transparency and post-market monitoring requirements?

• AI-based clinical decision support systems must satisfy both vertical AI Act safety obligations and horizontal interoperability requirements under frameworks such as MyHealth@EU, a dual-compliance challenge that many US-origin tools have not addressed.

Data residency and GDPR

• Where is patient data processed and stored? EU data residency is not optional for many NHS trusts and EU health systems.

• Does the tool's data processing agreement reflect GDPR obligations, including data minimisation and purpose limitation?

Guideline alignment

• Does the tool reference guidelines relevant to European clinical practice, including NICE, SIGN, AWMF, HAS, or national equivalents?

• Does the evidence base reflect European regulatory approvals and prescribing conventions, not just treatments approved by the US Food and Drug Administration?

Medical record system integration and workflow fit

• Does the tool integrate with the medical record systems used in European hospitals and primary care settings, or does it require clinicians to switch to a separate interface?

• Is it accessible via mobile, web, or embedded within existing clinical software?

Language and regional customisation

• For non-English-speaking EU member states, does the tool support clinical queries and responses in local languages?

• Does it accommodate local coding standards, including SNOMED CT and ICD-10?

Transparency and citation

• Can clinicians see the sources behind an answer? Citation-first design is increasingly considered a baseline for clinical trustworthiness, not a premium feature.

OpenEvidence: what it offers and where it falls short for European use

OpenEvidence is an AI-powered clinical search platform that generates answers to clinical questions grounded in peer-reviewed medical literature. Its design prioritises speed and source transparency, surfacing relevant studies alongside synthesised answers. For US clinicians, it has become a routine point-of-care tool.

For European clinicians, the picture is more complicated. The platform's evidence base is predominantly US-centric, reflecting FDA approvals, American clinical guidelines, and US prescribing norms. This creates a meaningful mismatch for clinicians working within NICE guidance, AWMF recommendations, or national formularies that differ from US practice.

More fundamentally, OpenEvidence is not currently available in the EU or UK. The withdrawal reflects genuine regulatory complexity: the EU AI Act's classification of clinical AI as high risk imposes obligations, including human oversight requirements, conformity assessments, and post-market monitoring, that require substantial investment to meet. This is not a temporary gap likely to close quickly. Platforms that are already MHRA-registered or CE-marked have a regulatory head start that takes years and substantial investment to replicate.

Tandem Health: clinical decision support within the consultation

Tandem Health is a European-built AI medical assistant that approaches clinical decision support differently from standalone query tools. Rather than functioning as a separate search interface, Tandem operates within the clinical consultation itself, using ambient voice technology (AVT) to listen to the encounter, generate structured clinical notes in real time, and surface AI-assisted clinical reasoning within the workflow rather than alongside it.

This embedded approach has practical implications for cognitive load, the mental effort required to manage competing tasks during a consultation. Clinicians using a standalone decision support tool must context-switch: formulate a query, leave the consultation interface, interpret a result, and return to the patient. Tandem's design reduces that friction by integrating documentation and decision support into a single clinical moment.

Tandem holds CE marking and MHRA compliance, and operates under ISO 27001 certification with GDPR-compliant data handling, meeting the data security and privacy standards that European health systems require. Its clinical decision support is now also EU MDR Class IIa certified. For clinicians in primary care or outpatient settings who want decision support that does not add a separate tool to their workflow, this integration model represents a meaningfully different value proposition from query-based platforms.

Other European or Europe-compatible clinical decision support platforms

Several platforms are worth considering for European clinicians evaluating their options. Each has a distinct primary use case and regulatory posture.

iatroX

iatroX is a UK-based clinical AI search platform with MHRA registration, drawing on a corpus that includes NICE, Clinical Knowledge Summaries (CKS), and SIGN guidelines. It offers citation-first answers and an integrated continuing professional development (CPD) system. It is available free to individual clinicians and is designed specifically for UK-accepted guidance rather than open-web retrieval.

Medwise AI

Medwise AI is positioned as an NHS-focused search tool with local and national guideline integration, backed by Innovate UK funding and listed on the NHS marketplace. A prospective pilot study at a Welsh NHS hospital found that the platform can collate local clinical guidelines, breaking them into retrievable chunks using natural language processing (NLP) to deliver concise answers to clinicians' questions. It is deployed at NHS Trust enterprise level rather than to individual clinicians.

AMBOSS AI Mode

AMBOSS AI Mode offers natural language search across AMBOSS's curated clinical content library, with inline citations and traceable source links. AMBOSS is German in origin, with a strong education heritage and structured learning pathways. Its curated-content model, where evidence is selected and maintained by physician editors, provides a different trust approach from tools that retrieve answers from open guideline corpora.

Isabel DDx Companion

Isabel DDx Companion is a differential diagnosis support system with UK origins, covering more than 10,000 conditions. It accepts signs and symptoms in natural language and returns a ranked list of potential diagnoses. It is not a guideline engine; its function is to surface possibilities a clinician might have missed, not to direct management. It is available to individual clinicians and institutions.

Vera Health

Vera Health is HIPAA and GDPR compliant and available worldwide without regional restrictions. It has no NPI or regional access requirement, making it accessible to European clinicians. In a recent evaluation, Vera Health scored higher than OpenEvidence on source-linking and workflow coverage.

UpToDate Expert AI

UpToDate has incorporated AI-enhanced features including a conversational assistant grounded in the UpToDate content library, with clickable citations and step-by-step clinical reasoning. These features are accessible to European clinicians via institutional subscriptions and NHS OpenAthens access. Its editorial depth is a key differentiator.

DynaMed

DynaMed covers more than 3,400 clinical topics with daily updates and explicit levels of evidence. It has received multiple Best in KLAS awards for Clinical Decision Support and is available to European institutions.

Pathway MD

Pathway MD is a Canadian mobile-first tool that presents clinical guidelines as decision algorithms. It is available free or on a freemium basis and covers a range of international guidelines.

How clinical decision support fits into the broader clinical workflow

The most effective clinical decision support tools connect to the broader architecture of clinical work, including documentation, referral, triage, medical record data entry, and coding, rather than sitting as a separate lookup function.

A cross-sectional survey across six European countries and the US found that around half of healthcare professionals reported using clinical decision support systems, with significant variation across countries. Participants reported using these tools mainly for diagnostic purposes or guideline implementation, but identified prognostic tools as the area of greatest unmet need. This suggests the current generation of query-based platforms addresses only part of what clinicians actually require.

Research on automated guideline adherence monitoring points toward an architectural direction that European health systems are increasingly pursuing: integrating guideline recommendations with real-time clinical data from medical record systems using Fast Healthcare Interoperability Resources (FHIR)-based interoperability standards. This approach moves clinical decision support from a reactive lookup tool to a proactive, patient-specific system, though implementation at scale remains complex.

The evidence on whether decision support tools change clinical behaviour is more nuanced than vendors typically acknowledge. A cluster randomised trial of the European Society of Radiology iGuide clinical decision support system across three German university hospitals found that the system did not produce a statistically significant reduction in inappropriate imaging requests compared with control departments. This is a useful reminder that the presence of a clinical decision support tool does not automatically translate into changed clinical behaviour. Implementation context, workflow integration, and clinician engagement all matter.

Key questions to ask before adopting a clinical decision support tool in Europe

For clinicians and clinical leads evaluating a new platform, the following questions provide a practical due diligence framework. For UK clinicians and healthcare organisations, the bar for safe adoption is set by the Digital Technology Assessment Criteria (DTAC) and NICE Evidence Standards Framework, not by marketing claims or US adoption metrics.

On data and regulatory compliance:

• Where is patient data processed and stored? Is EU or UK data residency guaranteed?

• Is the tool registered as a medical device under EU MDR or MHRA frameworks?

• Has it been assessed under the EU AI Act's high-risk AI provisions?

• Does it meet DTAC requirements for NHS use?

On clinical content:

• Which clinical guidelines does the tool reference, and are they relevant to your national context?

• How frequently is the evidence base updated?

• Are answers fully cited, with traceable links to source material?

On workflow integration:

• Does the tool integrate with your medical record system, or does it require a separate interface?

• Is it accessible at the point of care, via mobile, web, or embedded?

• Does it support your local language and coding standards, including SNOMED CT and ICD-10?

On transparency and safety:

• Does the tool clearly communicate uncertainty or limitations in its answers?

• Is there a mechanism for flagging errors or outdated recommendations?

• Does the tool address real-world data quality challenges, including variability, bias, and distributional shift over time, that affect AI reliability in clinical settings?

The direction of travel: clinical decision support in European healthcare

The regulatory environment for clinical AI in Europe is becoming more structured. The EU AI Act's high-risk classification for clinical decision support systems means the bar for market entry, and for continued operation, is rising. AI-enabled clinical software that is not engineered with AI Act safeguards and MyHealth@EU interoperability requirements from the outset will face increasing friction as enforcement matures.

At the same time, nearly half of EU Member States have already created dedicated professional roles for AI and data science in health, and several countries have indicated plans to introduce or expand AI training programmes. The institutional infrastructure for clinical AI is being built, which means the tools that integrate well with that infrastructure will have a structural advantage over those that do not.

The shift toward AI-native operating systems in primary and secondary care, platforms that handle documentation, decision support, referral, and coding within a single workflow rather than as separate tools, reflects a broader recognition that admin burden and cognitive load are themselves clinical safety issues. European clinicians are not simply looking for a replacement for OpenEvidence. They are looking for tools that reduce friction, align with local governance, and support clinical reasoning without adding another system to manage. The platforms that will succeed in this market are those built with that constraint at their centre, not retrofitted to meet it.

Frequently asked questions

▶ Why did OpenEvidence withdraw from the UK and European markets?

OpenEvidence withdrew from the UK and European markets in early 2026, a move widely attributed to regulatory uncertainty surrounding the EU Artificial Intelligence Act. The Act classifies AI-based clinical decision support systems as high risk, triggering transparency, provenance, and robustness obligations that most US-built tools were not designed to meet. Meeting these requirements demands substantial investment, and the gap is unlikely to close quickly.

▶ What regulatory requirements must a clinical decision support tool meet for use in Europe?

A clinical decision support tool used in Europe must satisfy several overlapping frameworks. These include registration or certification under the EU Medical Device Regulation or, in the UK, MHRA registration. The tool must also meet the EU AI Act's high-risk AI provisions, covering human oversight, conformity assessments, and post-market monitoring. GDPR requires that patient data is processed and stored within the EU or UK, and NHS organisations in England must also meet the Digital Technology Assessment Criteria.

▶ What are the main limitations of OpenEvidence for European clinicians?

OpenEvidence's evidence base is predominantly US-centric, reflecting Food and Drug Administration approvals, American clinical guidelines, and US prescribing norms. This creates a meaningful mismatch for clinicians working within NICE guidance, AWMF recommendations, or national formularies that differ from US practice. Most fundamentally, the platform is not currently available in the EU or UK, so European clinicians cannot use it at all.

▶ Which clinical decision support tools are available to European clinicians as alternatives to OpenEvidence?

Several platforms are available. iatroX is a UK-based tool with MHRA registration, drawing on NICE, Clinical Knowledge Summaries, and SIGN guidelines. Medwise AI integrates local and national NHS guidelines and is deployed at NHS Trust level. AMBOSS AI Mode offers natural language search across curated clinical content with German origins. Isabel DDx Companion supports differential diagnosis across more than 10,000 conditions. Vera Health is GDPR compliant and available without regional restrictions. UpToDate and DynaMed are both accessible to European institutions via subscription. Each has a distinct regulatory posture and primary use case.

▶ How does Tandem Health differ from standalone clinical decision support query tools?

Tandem Health is a European-built AI medical assistant that operates within the clinical consultation itself, using ambient voice technology to listen to the encounter, generate structured clinical notes in real time, and surface AI-assisted clinical reasoning within the workflow. Standalone query tools require clinicians to context-switch: formulate a query, leave the consultation interface, interpret a result, and return to the patient. Tandem's embedded approach reduces that friction. It holds CE marking and MHRA compliance, and operates under ISO 27001 certification with GDPR-compliant data handling.

▶ Does clinical decision support technology actually change clinical behaviour?

The evidence is more nuanced than vendors typically acknowledge. A cluster randomised trial of the European Society of Radiology iGuide clinical decision support system across three German university hospitals found that the system did not produce a statistically significant reduction in inappropriate imaging requests compared with control departments. Implementation context, workflow integration, and clinician engagement all influence whether a tool changes practice, not simply its presence.

▶ What questions should a European clinician or clinical lead ask before adopting a clinical decision support tool?

Key questions cover three areas. On data and compliance: where is patient data processed and stored, is EU or UK data residency guaranteed, and is the tool registered as a medical device? On clinical content: which guidelines does the tool reference, how frequently is the evidence base updated, and are answers fully cited with traceable source links? On workflow: does the tool integrate with your medical record system, is it accessible at the point of care, and does it support your local language and coding standards, including SNOMED CT and ICD-10?

▶ How important is guideline alignment for clinical decision support tools used in Europe?

Guideline alignment is a practical requirement, not a preference. European clinicians work within frameworks such as NICE, SIGN, AWMF, and HAS, which can differ substantially from US clinical guidelines and FDA-approved treatments. A tool that surfaces recommendations based on US prescribing norms may give clinicians guidance that does not apply to their national formulary or regulatory context. The evidence base a tool draws on should reflect the clinical environment in which it is used.

▶ What does the EU AI Act mean for clinical AI tools in practice?

The EU AI Act classifies AI-based clinical decision support systems as high risk. This triggers obligations including transparency requirements, conformity assessments, human oversight mechanisms, and post-market monitoring. Tools must also satisfy interoperability requirements under frameworks such as MyHealth@EU. AI-enabled clinical software that is not built with these safeguards from the outset will face increasing friction as enforcement matures. Platforms that already hold CE marking or MHRA registration have a regulatory head start that takes years and substantial investment to replicate.

▶ What unmet needs do European clinicians report around clinical decision support?

A cross-sectional survey across six European countries and the US found that around half of healthcare professionals reported using clinical decision support systems, mainly for diagnostic purposes or guideline implementation. Participants identified prognostic tools as the area of greatest unmet need. This suggests the current generation of query-based platforms addresses only part of what clinicians actually require, and points toward a broader need for tools that integrate with real-time clinical data rather than functioning as reactive lookup systems.