What NHS staff need from evidence services: findings from our 2026 user research

It seems like only the blink of an eye ago that Alan Fricker was launched across the NHS in England.

However, the process to reprocure the national discovery tool and accompanying infrastructure is underway.

To support this we carried out new research to improve our understanding of the evidence tasks our users are trying to complete and how they approach these tasks.

Our 2026 user research builds on the needs originally identified in the National Discovery Service research delivered by Lagom Strategy in 2019. The aim was not simply to ask users which products or features they preferred. It was to understand what NHS staff need to achieve when finding, accessing, using and sharing evidence, and to identify where their behaviours and expectations have changed. This post provides an overview of the report and the illustrative quotes in the full version are well worth a look.

How the research was carried out

The study had two stages.

First, members of the national Knowledge and Library Services team conducted 32 semi-structured online interviews. Participants came from a range of professional groups and settings, including acute, community and mental health trusts, primary care, commissioning support and national organisations. Interviews explored participants' roles, evidence needs, information-seeking behaviour, tools, devices, access problems, updating practices, use of library services and experience of generative AI.

The interviews were recorded and automatically transcribed. An experiment with AI "summarisation" proved unhelpful. While the AI summary recorded the topics and some degree of sentiment it flattened all the nuance in participants responses and stripped out much of the depth. Researchers documented observations using a common interview framework with quotes included where they were felt helpful. The material was thematically analysed by the lead researcher, supported by a workshop in which the interviewers compared their impressions and identified recurring or particularly striking themes.

The second stage was a validation survey. Twenty user-need statements were developed, most matched those used at the same stage of the Lagom research but with additions covering structured searching, access to internal and external evidence, and AI-supported evidence summaries. Respondents scored each statement from zero, “no need”, to four, “definitely need”. The survey received 473 responses from staff across at least 73 organisations.

There were limitations to our sample. Recruitment relied heavily on Knowledge and Library Services networks and in particular emails to NHS OpenAthens registered users. The sample at both stages of the research therefore contains more active users of evidence and library services than would be expected in a random sample of the NHS workforce.

The fundamentals have not changed

NHS staff are heavy users of evidence and their requirements extend beyond journal articles. They need evidence produced both within and outside their own organisations. Evidence related work is often fitted around other duties, undertaken during travel or completed outside normal working hours.

Across the validation survey, 13 of the 20 statements were rated as representing a very high need and five as a high need. 10 needs were rated as very high across all the role groupings examined.

1. Search from any device including desktops and mobiles
2. Search from any location, including outside the workplace
3. Find the relevant guideline for a condition or procedure
4. Follow a link straight to the full text of an article from search results
5. Filter a broad search to find the most relevant results
6. Access internal and external evidence and reports (new)
7. Receive the latest updates on a topic
8. Request a book or journal that isn't freely available
9. Store copies of articles I’d like to keep
10. Share a link to a piece of information with a group of colleagues

Access must be quick and dependable

"I actually only really get what I want probably about a third of the time because most of the time it is kind of paywalled."

Participants described multiple barriers from authentication, broken links, unavailable subscriptions and uncertainty about the correct route to a resource. Some described complex workarounds including trying via multiple organisational affiliations right through to contacting authors directly. This seems a missed opportunity where a request to a local service might have been simpler and quicker.

Crucially, inconvenience leads to abandonment. Participants described moving to another article if they experience barriers. The issue is therefore not merely whether content can eventually be obtained it can be a question of momentum in the window of opportunity they have.

This was reinforced by the survey comments, where full-text access and removal of paywalls formed the most common additional theme.

Relevance matters

"On the wards you haven't got time to go to the loo, never mind researching evidence."

Google, Google Scholar, PubMed and publisher websites remain important starting points. Participants valued convenience, but they also reported difficulties with broad or poorly ranked results. Relevance, clear presentation, useful filters and minimising the number of clicks were recurring priorities.

This has direct implications for the Hub. A large result set does not in itself meet user needs. Users want help getting from a broad question to a manageable and credible set of results, without requiring specialist knowledge of search syntax. This speaks to some of the support that new interface options might offer.

Mobile is not the whole story

"Using my phone because it's, again, quicker. It's the speed that makes the difference and the speed of access."

The laptop appears to have become the main workspace for concentrated evidence work. Mobile phones are important where speed, convenience or working on the move matters, and can provide an alternative where NHS devices or networks are restrictive. However, participants still found phones unsuitable for reading long articles.  Our sample also included some support for the value of printing for some forms of close reading and annotation.

The requirement is therefore not simply “mobile first”. It is for a consistent experience that works across devices, that is compatible with assistive technology and where is it practical to work between different devices / settings.

AI is of interest, but trust is paramount

"It's very good at producing very authoritative and believable responses. But I think in medicine, you need to know that it's correct."

Generative AI has become part of the evidence environment, although adoption in our sample was uneven. Interviewees valued its speed and ability to provide "answers".  There was appreciation of how it could help provide a starting structure. At the same time, they raised concerns about confident but incorrect answers, invented references, lack of transparency and responses biased towards the US healthcare context.

"I find that it's a challenge because... I haven't read this source material myself."

An AI-generated evidence summary was not among the highest-rated needs in the validation survey. The trust issue, along with a more library literate sample than average, may be a factor in this. Free-text comments nevertheless showed demand for concise, evidence-based summaries, particularly where sources are visible, limitations are clear and outputs can be adapted for different audiences. There was some indication of respondents wanting control to turn AI features off.

The message is not that users reject AI. Rather, AI functionality must earn trust. In a health care setting, provenance, accuracy and transparency matter.

Human support remains sought after

"Librarians are invaluable in helping search literature, accessing grey literature and helping with search strategies,... I would strongly advocate and recommend librarians being recognised as a resource that is needed."

The research also demonstrates the continuing value of Knowledge and Library Services staff. Respondents valued help with literature searching, grey literature, access to hard to source articles and guidance on what resources were available. Some specifically asked for brief, accessible training and opportunities to talk to someone directly.

Highlighting the visibility of local support remains a key deliverable for any future version of the Hub.

What next?

The research suggests a clear set of priorities for our ongoing development work and future procurement:

1. Remove friction between discovery and access, particularly around authentication and full text
2. Improve relevance and filtering, not simply the range / number of sources searched
3. Push for more internal and local sources in a coherent search experience
4. Support access across devices and locations
5. Make library support and document supply visible at the point of need
6. Use AI cautiously, with clear sources, transparent limitations and meaningful user control
7. Preserve continuity, recognising the cognitive cost of fragmented brands, changing routes and duplicated services

The tools available to NHS staff have changed since 2019, most visibly through the arrival of generative AI. Their underlying needs have changed much less. Staff still want relevant, trusted evidence, available quickly and with as little friction as possible. That should remain the central test for future development and procurement.