☰ CCIO handbook contents

Chapter 10


Big Debate: Common Standards

Let’s do it like the internet!
By: Ewan Davis

Standards are central to interoperability and yet the way we do them in healthcare is a barrier to innovation and progress. The mantra of “ruthless standardisation” of the National Programme for IT got us nowhere and current formal standard setting process is not fit for purpose.

Yet we have a towering example, with the internet, of how a loose, user-driven approach to setting standards can deliver practical interoperability and transformational change. So why don’t we do it like the internet in healthcare?

Different strokes for different folks

Before trying to answer this question - and to persuade you that we should follow the internet approach - I want to differentiate between two broad classes of standards: technical standards and content standards.

Technical standards are about the bits and bytes. They describe how data can be communicated between systems, but say little or nothing about the meaning of the data.

Such standards are usually generic (in other words, they are not specific to healthcare) and in general clinicians don’t need to know much about them. They include ASCII, XML, MIME, JSON, SMTP, TCP/IP, etc, etc ad nauseam – look them up if you are interested; but it’s probably best not to bother, as the details can safely be left to the techies.

Content standards define what we want to communicate and how we want to represent it. The payload of content standards is called “clinical content” and can include any piece of clinical information or knowledge.

The experts in clinical content are the frontline clinicians working in the clinical domain represented by the content – not the informaticians, whose job is to make the content machine processable as far as this might be useful, or the engineers, whose job is to make the technology secure, reliable and performant.

It is essential that frontline clinicians assert their supremacy and leadership with regard to clinical content and are not bamboozled or dictated to by the high priests of health informatics.

My observation is that if you don’t understand what these people are talking about then it’s probably not worth trying to work it out. They should be able to articulate what clinicians need to know in a way clinicians can understand.

Clinical content includes things such as the definition of the elements of health records (that is: what do we mean by a diagnoses, medication, allergy, procedure and so forth, what properties do these concepts have and how do we represent them?) and descriptions of clinical knowledge, processes, pathways and decision making.

Standards in this area include SNOMED, OpenEHR, ProForma, PRSB Headings, HL7 – all things I would suggest it’s worth looking up if you don’t already know about them.

Why we should do it like the internet

So how do we best develop clinical content standards? Well my answer is simple: “like the internet”. That means: not like the way we have historically tried to do it, with formal processes like those followed by national and international standards bodies (BSI, CEN and ISO) and the NHS (CRIR, STEP, ISB and now SCCI).

So how does the internet do it? It does it by creating an environment in which it is easier to use available standards than not to use them, and by allowing anyone who wants to contribute to the standards development process to contribute to it.

It does it without a central approvals process by letting users decide what works for them and what doesn’t, by allowing variation and messiness, and by recognising that there are no permanent or perfect solutions, just things that are good enough for now.

The name for Internet technical standards says all you need to know about them – they are called “Requests for Comments”.

I believe one of the best examples – and one we should follow in healthcare – is that of Wikipedia. Here content is created and curated by self-selecting experts who feel they have sufficient expertise in a particular area to create a Wikipedia entry for it.

Anybody can create or change an entry. While this can result in some rubbish, if it does – and if the content matters to a significant number of people – it will be improved, with community pressure will suppressing malicious or ill-informed views.

The nature of the medium allows sincerely held but incompatible views to develop independently, allowing the user to choose their truth. Wikipedia has only lightweight governance structures which include some basic rules and mechanisms to coordinate related areas of content and deal with the few individuals of determined, malicious intent.

Let’s compare the approach of the internet and that historically used by the NHS.

  Internet NHS Impact of the NHS’s approach
Speed New content can be created and published in hours Approval and publication of new content takes months to years People forced to use bespoke non-interoperable approaches to meet immediate priorities
Authority to publish When the author(s) believes it’s fit for publication National or international standards/professional body As above, plus a tendency for politics and bureaucracy not domain expertise to determine decisions
Openness Standards generally open source and freely available to all NHS standards are freely available but those from standards and professional bodies may require payment to view or use A copy of a BSI/CEN/ISO standard typically costs a few hundred pounds Small and medium sized enterprises in particular remain ignorant of BSI/CEN/ISO standards because they are hidden behind paywalls and they can’t afford cost of finding out if a standard is relevant.
Licensing cost slows or precludes use of relevant standards
Ease of engagement Easy to engage in just those areas where a user has expertise and interest.
All parties engage on a equal basis
Heavy time and learning commitment and sometimes onerous confidentiality requirements to participate.
NHS bodies often exclude vendors or treat them as second class participants
Process gets left to bureaucrats and professional standards makers, often remote from real-world requirements and implementation issues

So how do we apply an internet style approach to the standards used in healthcare? For technical standards it’s easy – use those that the internet uses. Clinicians don’t need to get involved and only limited engagement in the standards process by techies from the health domain is required.

Content standards are more challenging and require active engagement from frontline clinicians, active support from NHS and professional bodies, vendors and domain techies.

We need to identify or create the governance structures, organisations and tools needed to support frontline clinical engagement in the creation and curation of clinical content and mechanisms, ensuring this is coordinated with international efforts.

We also need to identify funding and business models to make sure that the outputs are freely available to all who might benefit from them.

My view is that the central requirement is the creation of a fully open source and shareable information model that is capable of describing all of the elements of clinical content that we need to share.

There are existing methodologies and a considerable body of existing work that we can draw on. What we need is an initiative to make it happen. This is something in which the CCIO Leaders Network could play an important role.

About the author:Ewan Davis is director of Woodcote Consulting. A health informatics strategist, he previously worked at McKesson and BT Health. For more vies on standards, and about the role of open source software in healthcare, check out his blog: http://www.woodcote-consulting.com/lets-do-it-like-the-internet/

Let’s not do it like the internet!
By: Professor Iain Carpenter

This piece is about why I believe standards are needed in health informatics - how we record, access, process and transmit clinical information in electronic record systems.

It is about why the human readable end of the process needs to be agreed by patients and care professionals – and the technical bits agreed with techies.

It is about why it is not good enough to leave these matters to the whims of the moment, the internet and the enthusiast. Nor is it appropriate to wait until everything is defined before we start.

Context matters

I am a clinician and I want to be sure that clinical information I access is accurate and up to date. That particularly applies to clinical information related to my patients; and I know my patients want the same thing (because they have told me so, again and again and again).

I know that when clinical information is missing, incorrect, or out of date, things can go terribly wrong. Fitness to practice reports of the General Medical Council, Royal College of Nursing and defence organisations, reports of the Health Ombudsman and in the media, and stories ‘from my neighbour’ all say the same thing.

Here’s an example - I can talk with my patient about their risk of myocardial infarct. I record that in the notes, and when we read the notes, the patient my colleagues and I know that there is a risk of MI.

It is not a current diagnosis, previous medical history (PMH) or family history (FH) of MI. We will know that because of where and how it is recorded – risk of MI.

The context of MI, ‘risk of’ rather than diagnosis, PMH or FH is critically important. For that information about the clinical data item ‘MI’ to be exchanged between one computer and another, with no ambiguity, then the context must be defined in the same way, or things can go wrong.

Clinical information is complex, and its precise meaning is extremely dependent on context. Context is an essential component of “semantic interoperability” for clinical information, “the ability of computer systems to exchange data with unambiguous, shared meaning”

Here is a second example. A very depressed man in the US was sent for a CT scan to exclude a cerebral tumour – he was told he didn’t have one.

He could not understand why he was turned down for medical insurance six months later, and was devastated to hear that this was because he had a cerebral tumour. The insurance company had accessed his medical record, with his permission. Its clinical information system did not have a way of recording ‘differential diagnosis’ and had instead recorded ‘diagnosis’ of cerebral tumour.

Both examples demonstrate the crucial importance of precise information to a patient’s circumstances. Without standards for defining context, we will struggle to develop electronic record systems that allow patients to view their clinical records from any care provider on any device, wherever they are.

In addition, we need these standards so that clinical data can be aggregated for evaluating quality and outcome of care, for epidemiology, for post marketing surveillance of pharmaceuticals.

Ten years ago the Health Informatics Unit of the Royal College of Physicians led a project that developed standard headings for clinical records – with the participation and contributions of thousands of patients and clinicians.

These headings have been revisited and published by the Academy of Medical Royal Colleges with the endorsement of over 50 professional organisations, and adopted by the Professional Record Standards Body (PRSB).

On their own they are not sufficient – they cannot be directly, unambiguously built into computer systems - but they are necessary.

We are in an open dialogue, always advancing but based on common definitions – standards. Committed health informaticians link with clinicians and patients who use the terms and system vendors who have to make them work in the computers, in a three way dialogue.

None of us can live solely in our own space and believe that what we produce will work… because at some critical point it won’t.

About the author: Professor Carpenter is executive chair of the Professional Records Standards body and associate director of the Health Informatics Unit at the Royal College of Physicians. He was previously a consultant geriatrician, and is Emeritus Professor of Human Ageing at the University of Kent.

Top

Sorry, The CCIO handbook is not compatible with your browser

The CCIO handbook can be viewed on all modern browsers and on Internet Explorer version 9+.

Suggested browsers

Chrome

FireFox

Opera

Safari