Using AECMA 1000D/ATA 2100 data-sets to generate Class IV IETM's   Table of contents   Indexes   Case Study: Boeing Intelligent Graphics for Airplane Operations and Maintenance
 

A Healthy Revolution
 Andrew   Roberts
  Consultant Orthopaedic Surgeon
  Robert Jones & Agnes Hunt Orthopaedic Hospital  Oswestry
Shropshire   West Midlands  United Kingdom  SY10 7AG
Phone: +44 1691 4045730
Email: andrew.roberts@rjahoh-tr.wmids.nhs.uk Web: http://www.keele.ac.uk/depts/rjah/sgml/sgml.htm
 
Biographical notice:
 
Andrew Roberts has been exploring the use of SGML for patient records for the past six years. A pilot project conducted in 1995/6 demonstrated the feasibility of this approach and he is now actively involved in implementing a full working patient record solution at his hospital
 
ABSTRACT:
 
As the Millennium was drawing to a close, it became apparent that the NHS  (British National Health Service) had to change to meet the challenges ahead. With an ageing population and increasing technological costs, the remaining area of clinical practice where reform had not reached was addressed in the strategy document "Information for Health"(http://www1c.btwebworld.com/imt4nhs/ ).
 
At a time when information was becoming increasingly valuable, the Internal Market reforms were implemented by Margaret Thatcher's government in the late 80's and the first half of the 90's. A radical drive to improve efficiency led to a dramatic reduction in clerical staff. Information technology was directed towards transaction processing and supporting a newly introduced split between the purchasers and providers of health. Clinicians at this time, with a few fortunate exceptions, had to make do with paper records and all the limitations and frustrations which that brought. The Audit Commission, which acts for the British public purse to assure good financial practice and efficient use of resources, published a comprehensive indictment of paper records and indicated that migration to electronic records would address the problem - but how?
 
The NHS Information strategy document sets out clearly how patient centred information is to be held at the point of care as an EPR  (Electronic Patient Record) and as a series of summaries of care in an EHR  (Electronic Health Record) . Achieving a satisfactory EPR solution appears to have evaded most attempts for a number of reasons including the general lack of patient focus and an inevitable process based approach to patient records.
 
This paper explains how content management can address the agenda outlined in the "Information for Health" strategy document but the strategy is about information not technology. The power of SGML/XML to solve our needs does not mean that it will automatically act as the ball bearings for the revolution that must come. The NHS Information Special Health Authority hasnot endorsed the use of SGML/XML over any other form of technology. Having issued this caveat, how could SGML/XML solve the problems that lie ahead?
 

INFORMATION FOR HEALTH
White Paper
 
After the changes introduced by the Conservative Government a new Labour Government changed the emphasis from health run along business lines to a knowledge led healthcare system where efficiency savings accrue from improved care rather than competitive pressures. Information, in the new setting, is either patient centred or publicly accessible. Electronic patient records, the electronic healthcare record and the National Electronic Library for Health are the principle components which are amenable to expression in SGML/XML.
 EPR, Electronic Patient Record  
 

THE ELECTRONIC PATIENT RECORD (EPR)
 
The strategy document defines EPR as"the record of the periodic care provided mainly by one institution. Typically this will relate to the healthcare provided to a patient by an acute hospital. EPRs may also be held by other healthcare providers, for example, specialist units or mental health NHS Trusts". As such, content modeling in EPR focuses on existing documentation and processes with high levels of granularity to enable full support of the organisation's activities.
 
The SGML electronic patient record project that started in the Robert Jones & Agnes Hunt Orthopaedic Hospital, Oswestry has now been completed. Where possible, information which is digital is streamed directly into the patient centered EPR. This data comes principally from the core database at the hospital, known as the PAS  (Patient Administration System) , and from a number of legacy laboratory systems that deliver structured investigation results that are capable of being tagged.
 
At it's inception, the Oswestry project was seen as working in tandem with the PAS which acts as a master system containing demographic and coding data concerning all patients treated within the hospital. With the patient's hospital registration number acting as the single most important key field both in the PAS and SGML EPR systems, demographic data is automatically merged and tagged into patient documentation reducing the possibility of transcription errors.

The current system at Oswestry


A wide variety of information sources contribute to the EPR document repository

  
 
An agreed structure for each form of clinical interaction occurring within the hospital allows the construction of a simple prompt card for each document type. Voice recognition technology was not judged to have become sufficiently robust to allow direct verbal inputting and Dictaphones were used with medical secretaries entering the structured dictation into a correspondingly structured markup engine that replaced their word processors. By using the ability of the markup engine to produce new combinations of information many of the documents which were previously word processed were either completely or partially delivered automatically, saving approximately 25% in time compared with the word processor. Once secretaries were given the new engine, they were reluctant to return to the old way of working.
 
The PAS also gave a series of other feeds to the system that generated documents giving details of admission and discharge from hospital; operation dates; coding and future outpatient appointments. Each laboratory result was marked up and treated as a separate document with it's identity and time point entered in the registry so that the browsers could retrieve a registry report to display as a navigator tree along with the text and PAS events.
 
At the output side of the process, the developers were fully aware of the time pressures under which the clinicians were working and a sub second response was imperative. By feeding only the required documents to the browser, access times as well as network traffic were optimised. The initial view of the data consisted of the last event in the navigator tree as this was thought to be the most relevant to the following consultation but free access to all other events and items was through the navigator tree.
 
The browser was a custom built device that controlled access security; patient selection; multiple patient views and multiple event views. An important feature of the browser was that it should act as a single interface to all the available electronic data concerning the patient and have hooks to allow activation of any clinical support systems which might be available without the clinician having to re-(mis)enter patient identifiers. Actions such as signing off documents, adding alerts, initiating literature searches and plotting test results were included.

The "Doer's" browser showing an operation note


Text captured with a structured markup tool for SGML display

  
 
Whilst the possibility of using a commercial browser with Java applets delivering the necessary functions which the clinician needs might seem attractive we did not feel that this would allow the flexibility which could be gained with the use of a custom built healthcare browser. It remains to be seen whether the rapid development in XML technology will deliver software environments that are capable of behaving as reliably and robustly as a secure custom built device.
 
At Oswestry the system is currently (at the time of submission) working with live capture and delivery, in the Children's Orthopaedic Unit and the Professorial Orthopaedic Unit. Four consultants, six junior staff and five medical secretaries as well as nursing, physiotherapy and pharmacy staff have access to the system.
 
The system was primed with four years of legacy data which was in the form of word documents; a Q&A discharge summary database and database physiotherapy and pharmacy records. A total of just over 250,000 documents are now in the repository marked up with valid XML. By incorporating legacy data, the system is given "immediate value" and this improves take-up and user enthusiasm particularly when they realise that the prospectively collected data contains even greater granularity allowing more exact in context searching.

In context searching


An example of structured data being searched using the Inso Dynatext product

  
 EHR  
 

THE ELECTRONIC HEALTH RECORD (EHR)
 
Whilst most members of the public consider that healthcare is all about hospitals and high tech medicine, the truth is very different. Only a small proportion of the nation's healthcare budget is spent within hospitals, most going to support care in the community. Responsibility and financial control is very confused here, in the current British Health Service. Health Authorities fund all the services within their area. Family doctors are self-employed under contract to the Health Authority. Community Health Trusts fund Health Visitors, Community Physiotherapists; Community Midwives and the Community Health Service whilst the Mental Health Trusts fund Community Mental Health Nurses (http://www.bmj.com/cgi/content/full/317/7158/579 ). The axis of this constellation of care is about to shift as family doctors are to be brought together as PCTs  (Primary Care Trust) with responsibility for commissioning hospital services and other aspects of care for their patients.
 
The strategy document defines the EHR as:"The term Electronic Health Record (EHR) is used to describe the concept of a longitudinal record of patient's health and healthcare - from cradle to grave. It combines both the information about patient contacts with primary healthcare as well as subsets of information associated with the outcomes of periodic care held in the EPRs".
 
The ownership of the EHR is stated to be at the level of the PCTs :"The Electronic Health Record will inevitably be built and retained in primary care". This aspect of the strategy seems poorly thought out. The technological demands of providing a reliable record available in a secure fashion wherever and whenever the patient requires care are probably beyond the resources of PCTs. Furthermore, if the EHR is to achieve the widespread coverage which is envisaged, then a degree of uniformity will be required. This cannot be achieved without a degree of central control - which, as yet, has not been declared.

Data sources for the EHR


From the NHS White Paper - Information for Health

  
 
The functions of the EHR are stated to be both:"record transfer and, in anonymised and aggregated form, a vital aid to national and local health status analysis".

Uses for the EHR


From the NHS White Paper - Information for Health

  
 
The Latter function seems to be one where the benefits of document content markup will ensure that of all the technologies that could possibly be used, SGML/XML will succeed. Markup is essential is the records are to be effectively anonymised because not only does the demographic data have to be altered by omission and transformation but the name elements must be used to purge other free text fields of the patient's name and any name synonyms.
 
Once the data has been properly sanitised, either an XML aware object orientated database or an SGML repository can be employed to enable searching. Typically these searches are complex and require several lines of script to produce a useful result. Importantly, no search will be totally specific and sensitive. What can be said is that the degree of accuracy is better than where coded data is examined and that will give an accuracy which when applied to a gold standard data set gives a known positive and negative predictive value for the search target in question.
 
A pilot custom built browser has been constructed to enable the principles of dial in access to a core EHR repository. Rather than the ad hoc document structure of the EPR which depends upon scavenging the existing data sources for any relevant clinical information concerning a patient the EHR delivers synopses using a limited number of clinical headings to provide the main granularity of the record

EHR Browser


A simplified series of documents which summarise care episodes and act as a communication medium between health professionals caring for a patient.

  
Evidence
Information
Library
 

THE NATIONAL ELECTRONIC LIBRARY FOR HEALTH (NELH)
 
A considerable amount of information exists concerning health that thanks to the World Wide Web is freely accessible to the general population. The very openness of the web leads to a mixture of information varying from that which is authored to appraise patients and more complex material which is intended for clinicians.
 
The ease of publishing on the web makes the internet a soft target for publications which are insufficiently vigorous to merit publication in the peer reviewed medical literature and this has real dangers if viewed by the uncritical or untrained viewer. Even well trained clinicians are prone to accept poor quality information because the pressure to find information and deliver care is great. The strategy realises that both for clinicians and the general public, a reliable and well-validated repository of accurate information is a powerful force for health improvement promoting efficient clinical practice and empowering the population with knowledge about healthy living and illness.
 
The general population will access a National Electronic Library for Health through a service known as NHS Direct that provides health information using a variety of media including a helpline staffed by nurses. The clinician will have access to a different form of information that will be used to enable the practise of EBM  (Evidence Based Medicine) .
 
Highly structured EBM questions will be addressed with a wide variety of information resources to form CATs  (Critically Appraised Topic) so that a clinician implementing such a CAT can be sure that they are acting on the best research based information to provide efficient care. The structure of the EBM questions and the CATs lends itself to expression in XML. A significant advantage of using mark-up for this process will be the reuse of CATs rather than duplicating what is certain to be a very considerable effort. The EBM side of NELH  (National Electronic Library of Health) activity seems certain to depend upon document content management both for manipulation as well as delivery.
 

CONCLUSION
 
Soon after the strategy was released, the excitement gave way to uncertainty. "Yes - we now know what information we must manage but how to do it?" So much of what must be dealt with is text. Traditionally healthcare computing retreated to laboratory and financial data whilst the bulk of the information clinicians want to see and analyse is text. There will be an expanding role for the traditional technologies however the text element seems certain to require significant management of a level that can only be provided by SGML/XML and all that laboratory data has to be published in an intelligent fashion. The future looks bright.
 
Acknowledgments
 
We have received considerable financial support from the Electronic Patient Record Program of the National Health Service Management Executive. Without their support, this presentation would have been speculative rather than factual.
Using AECMA 1000D/ATA 2100 data-sets to generate Class IV IETM's   Table of contents   Indexes   Case Study: Boeing Intelligent Graphics for Airplane Operations and Maintenance