The NLH Enterprise Architecture – looking to the future

For a couple of years now I have had a copy of “An Enterprise Architecture for the National Library for Health: Direction of travel and deliverables” sat in my inbox.  At the time I was preparing a report considering HDAS (the NHS in England interface for searching literature databases) and I wanted to remind myself of the content of this dimly remembered document.  Linda Ferguson kindly dug it out (on a site since dead) and it has been sat nagging me ever since.

The version of the EA above dates from 2006.  I confess I failed to grasp the scope of the vision it represented at the time.  The language is by nature technical but the ambition is very clear.  A number of initiatives now under way in the NHS in England could be plucked directly from this document and would certainly be much easier to deliver if we had gone further down some of the paths it suggests.

The document lays out a plan for delivery of NHS Library web based services.  At the core is the need for “a set of interoperable, networked services that conform to appropriate open standards”.  This would be supported by various things such as shared schemas for meta data and a central registry for API specifications.

I want to consider how the implications anticipated at the time have worked out and where things might be going in the future.

There were several implications identified for national services


A coherent and integrated user journey is desired. Presentation layers, what the user sees as a web page and how results are presented, will be separate from content and services and owned and built by the NHS.

Procurement will focus on content and the necessary APIs to integrate content into the discovery and current awareness processes. Increasingly, we do not wish to purchase content locked into any single portal.

A core search service will index all NHS content. It too will have a SOA, providing the basis for search pages. It will integrate with related services such as link resolvers

An NHS resolver service will be a key component in the delivery process. The NHS will wish to procure and own a resolver solution as a managed service.

An NHS library– wide Access Management System is being procured. Use of this system will be mandated for information suppliers. It will be SAML compliant.

Much of this has come to pass though perhaps without the core search service.

HDAS has reasonably successfully allowed for changes to the suppliers of content (databases) without massively impacting the experience of searching for the end user.  The varying API offered by suppliers have not fully supported the consistent search experience desired and there have been performance issues.  What has not happened perhaps is the ongoing integration of other services such as document supply and support at the point of need into HDAS.

We have seen the procurement and integration into HDAS of different link resolving solutions.  OpenAthens has been a long standing partner for access management.

Looking to the future work is underway to deliver an NHS England wide discovery solution and how well this maintains control over the web page and presentation of results will be interesting.  This could potentially be the “core search service”?  NHS Evidence already does this job for some categories of materials but stays away from the literature databases that would swamp the materials it aims to present.

A missed opportunity was the investment to create an NHS England wide Library Management System based in one of the Open Source solutions.  A small central team could have administered and developed a tailored approach that would have matched some of the ambititions of the EA.  I suspect the overall cost over the past decade would have been significantly lower and the opportunities for creating a platform for services greater.

There were also implications flagged for local services

local e-content, whether procured or NHS generated can, by adhering to EA principles, be integrated with national content, either within NLH or within other portals.

New services can be built up around this technology. For example, local current awareness and alerting services can be integrated with national services to provide the user with one way of getting knowledge updates

A single NHS library-wide Access Management System provides web Single Sign-on linking library services to their user base and will provide a bridge to NHS SSO services, opening up library service to non-library users.

Generally we have been happily plugging in locally procured content into national systems.  A gap has been around a solution for ebooks and this will need to be addressed in any new discovery layer as this format grows in importance.  The ability to integrate local content will depend on standards and considering these might be an early priority (as fixing them later will be trickier).

Recent revisions to HLISD will hopefully have maintained the commitment to the important location and service information being available via API to build other service offers.  The wide adoption of KnowledgeShare raises questions of how this (or an equivalent) might be integrated into a future national digital service.

In an ideal world we would have single sign on using peoples Trust logins – any additional login (even one as familiar as OpenAthens) is an unwanted barrier so an NHS SSO is the right ultimate target.

So quite a lot progressed and quite a lot left to do.  As the NHS in England moves towards the procurement of a new discovery tool it feels to me more critical than ever that we maintain the drive of the NLH Enterprise Architecture for the delivery of “a set of interoperable, networked services that conform to appropriate open standards”.  What I would like to see more of is the role of the person supporting at the point of need within that networked service.

These are my views based on my involvement with various aspects of the health libraries system at different points.  I am very happy to be corrected on points of accuracy – and challenged on matters of opinion!

Tongue tie and the supplementary concepts

Delivering tailored literature search training for a departmental group is always a great opportunity for learning new things.  Today I helped a multiprofessional group who work in a tongue tie clinic.

This is the (cleaned up) strategy we ended up creating as part of the session:

  1. MEDLINE; (tongue AND tie).ti,ab; 160 results.
  2. MEDLINE; tonguetie.ti,ab; 6 results.
  3. MEDLINE; ankyloglossia.ti,ab; 290 results.
  4. MEDLINE; “tongue tie”.ti,ab; 157 results.
  5. MEDLINE; 1 OR 2 OR 3 OR 4; 398 results.
  6. MEDLINE; breastfeeding.ti,ab; 14979 results.
  7. MEDLINE; (breast AND feeding).ti,ab; 14567 results.
  8. MEDLINE; 6 OR 7; 27717 results.
  10. MEDLINE; BREAST FEEDING/; 27307 results.
  11. MEDLINE; 5 OR 9; 1383 results.
  12. MEDLINE; 8 OR 10; 38085 results.
  13. MEDLINE; 11 AND 12; 109 results.

The keywords are a nice example with some extra results by dropping the space and a solid alternate term in ankyloglossia that has a big impact on total results.

More interesting again is moving to the suject headings.  There is no Tongue Tie/ and it is not immediately clear what to go to from the alternatives offered up.  When in doubt try Pubmed and a tongue tie search in Pubmed returns the following search:

“Ankyloglossia”[Supplementary Concept] OR “Ankyloglossia”[All Fields] OR “tongue tie”[All Fields]

I had not met Supplementary Concept before (or if I have I forget).  When in doubt Google and that brough me to Mesh Record Types which explains these are for chemicals, drugs and rare diseases. Each is linked to a broader descriptor by the Heading Mapped to field (HM).

Sadly HDAS does not include HM in the “more fields” option and I cannot see a likely candidate in the list I tried HW but that came up blank.

For the purposes of my session I used the MeSH browser to look it up and found

Heading Mapped to *Mouth Abnormalities

I did not note the * at the time and wonder if I should have restricted to major terms but given the small numbers of articles involved I think it was fair to go with the whole heading.

Finally I also had a look in Cochrane where there is a protocol Frenotomy for tonguetie in newborn infants (which does not come up if your search for ankyloglossia incidentally). The protocol claims there is Tongue Tie in MeSH – it is an Entry Term but it won’t map to it so not sure myself.

Welcome anyones ideas on how to explain this to people!