Cross-boundary communication: the development of structured and computerised formats for admission and discharge documentation and an audit of their use
Rosemary Keele1, Brian Payne2, Ged Keele3
1Project Co-ordinator
2Consultant Geriatrician, Director, Medicine for the Elderly, West Norwich Hospital, Norfolk and Norwich Hospital Health Care Trust
3General Practitioner, Dereham, Norfolk
Overview and background
The Department of Medicine for the Elderly in Norwich set up a cross-boundary Information System (IS) development project with local general practitioners in 1987. Early work on the Norwich Geriatric Information Project (NGIP) concentrated on analysis of the hospital record to establish consistent formats for manual use. An initial computerised prototype of a ward-based discharge summary was developed using Version 2 Read Codes. This work provided the basis for audit work on referral and discharge communication in acute admissions of elderly patients. The audit was supported by the NHS Management Executive programme of audits in Primary Care during 1993. The first two sections of this paper describe the initial audits of communication documents and reaudits following the introduction of changes in procedures. The final section summarises subsequent IS development on two further projects which have produced enhancements of the prototype environment. These prototypes are currently the subject of further trial work in hospital and general practice.
Hospital based audit work
Method
The audit work was undertaken within the framework of the West Norwich Department of Medicine for the Elderly audit programme. Initial audit survey work examined 250 discharge notes in the structured paper-based format that had been specifically developed for use in the care of the elderly on the NGIP Project. Standards for improved recording of information on these paper discharge documents were then set following internal Departmental meetings of consultant and junior staff. The reaudit survey was split into two parts: a second survey of 150 manual paper discharge note, and a trial of an enhanced prototype version of the computerised discharge note format in Protos Patient Care System software. The electronic discharge note data was input by the project co-ordinator in a "near ward setting" following perusal of the patient notes whilst the patient was on the ward.
Results of audit of missing information on paper discharge documents
Initial Follow up Significance
survey
(n=250) (n=150) Values
Patient Identification
Missing Name 3 2
(1%) (1%)
Address 8 8
(3%) (5%)
Date of Birth 25 (10%) 6 p<0.05
(4%)
Hospital Number 10 (4%) 4
(3%)
GP Details
Missing name 10 (4%) 6
(4%)
address 30 (12%) 6 p<0.01
(4%)
Episode details
Missing discharge date 8 2
(3%) (1%)
ward 8 6
(3%) (4%)
consultant 13 (5%) 6
(4%)
discharge destination 40 (16%) 14 p<0.05
(9%)
Clinical details
Missing diagnosis 5 3
(2%) (2%)
treatment statement 70 (28%) 35 (23%)
discharge drugs 5 2
(2%) (1%)
assessment 100 (40%) 18 (12%) p<0.01
Discharge plan
Missing OPD follow up 35 (14%) 21 (14%)
Service plans in 163 (65%) 50 (33%) p<0.01
community
Legibility problem 25 (10%) 20 (13%)
Conclusions from results of audit of paper based record
Audit of the computerised discharge note
Audit of the trial of a computerised discharge note cannot be compared directly with the audit of the paper-based documents. The project co-ordinator produced the computerised documents and was clearly intent on producing complete documents of high quality. By definition 100% legibility of documents was achieved and careful perusal of the notes ensured that other areas were filled in completely. Despite this there are difficulties with this particular approach to computerised discharge notes:
General practice audit
Method
The audit of general practice documents concentrated on emergency referral letters that accompanied patients on admission. This approach was taken because the hospital provided a focal point for letters to be reviewed by the project co-ordinator. A baseline audit was undertaken to establish levels of completeness of data on acute referral letters. Following this audit a standard manual form for use in emergency admissions by general practitioners was designed. This was circulated for comment to a panel of local general practitioners who had expressed an interest in improving communication at referral. All general practitioners in the Norwich Health District were then supplied with personalised A4-size structured referral forms and they were invited to use them when they admitted elderly patients to hospital. They were also sent a summary of the initial findings and an A5-size laminated checklist that summarised the desired content of a referral letter.
Results of base line survey of emergency GP referral letters
Referral letters in 250 acute admissions were studied. In 18 (7%) of these admissions there was no referral documentation, leaving 232 letters for analysis. The salient features of this analysis were:
Results of further survey work following distribution of structured letters to GPs
Following distribution of the structured referral letters, 125 acute admissions were studied. Only 2 of these did not have referral letters so 123 referral letters were available for the study. In this sample 22 letters were written on the structured form (18% of all letters studied). Computer printouts of problems and/or medication accompanied the letter in 14 admissions (11%).
Comparison between the structured and non-structured letters:
This comparison was made on preset criteria that were drawn up as result of the baseline survey work:
CRITERION Unstructured Structured letter Significance
letter
(completeness of data)
Date 88 (88%) 18 (82%)
Date of birth 91 (91%) 17 (95%)
Patient Address 90 (90%) 18 (100%)
Patient Post Code 8 (18%) 14 (64%) p<0.001
Recorded
Patient Tel. No. Recorded 8 (19%) 15 (68%) p<0.001
Hospital No. Recorded 52 (52%) 18 (82%) p<0.02
Admitting GP Recorded 79 (82%) 22 (100%) p<0.05
Admitting GP Address 97 (97%) 22 (100%)
GP Tel. No. 88 (89%) 22 (100%)
Registered GP Name 36 (36%) 22 (100%) p<0.001
Registered GP Address 31 (31%) 22 (100%) p<0.001
Problem/Diagnosis 97 (97%) 22 (100%)
Past Medical History 90 (90%) 22 (100%)
Presenting History 91 (91%) 22 (100%)
Examination findings 93 (93%) 22 (100%)
Medication 90 (90%) 21 (95%)
Treatment before 27 (27%) 20 (91%) p<0.001
admission
Functional assessment 32 (32%) 22 (100%) p<0.001
Social History 12 (12%) 21 (95%)
Conclusions from the audit
The results of a reaudit of referral letters showed that the structured referral form was acceptable to general practitioners. However, the method of disseminating the form did not promote a high uptake of use within the locality. The audit did produce a measurable improvement in the quality of information made available to the hospital at admission. From the hospital viewpoint, the improvements in information relating to functional assessment and social history were particularly important, because this information can sometimes be difficult to obtain in elderly patients who are confused.
Subsequent IS / IT development work
Overview of further project development
Further development work has been undertaken since the NHS Management Executive cross-boundary audit. The original concept of a shared record for hospital and general practice expanded to include all members of the health and social care team, and the project team began to work with academic staff at the School of Information Systems in the University of East Anglia (UEA,SIS).
The Computerised Integrated Record for the Care of the Elderly (CIRCE) ran from October 1993 to January 1995 and involved all the key agencies involved in the care of the elderly. It was supported by the East Anglia Regional Strategic Framework for Information and IMG Strategy A Branch. The Norfolk and Norwich Hospital Healthcare NHS Trust has now assumed ownership of the CIRCE Project and has agreed to support further work within the Trust provided that it is clearly separate from ongoing HISS implementation. The Trust sees a potential role for the project team in developing specifications for clinical system development in the IS/IT implementation phase of the new Norwich 2000 hospital. As part of this remit the work has been extended to include a Clinical Coding interface for the new CIRCE prototype workstation that has been developed in Microsoft Access. A joint project has now been set up between the Trust and Anglia and Oxford Regional Health Authority to support this work and some development in General Practice on Summary formats (the Problem- and Episode-based Recording System for Electronically Updating Summaries - PERSEUS - Project).
The general approach to Information System Development
The project consultants at UEA,SIS suggested that we use an evolutionary systems design approach to IS development rather than a structured design approach, for example, Structured Systems and Design Methodology (SSADM). This approach involves:
The reasoning behind this choice was three-fold. First, the project clearly needed a structured environment for further development to ensure that project products were more "open"; second, there was a need to be able to correct the course of development if there was a need to do so; and third, it fitted the project approach to user-centred prototyping development that had been developed on previous projects.
The CIRCE Project
The objectives of the CIRCE Project were:
Modelling the BCSR
The Basic Clinical and Social Record concept is built around the data structures required to support clinical communication. The original concept recognised that the primary need for clinicians to communicate across boundaries in a structured way should support a secondary requirement to produce administrative data to support the processes of NHS management. The principal aims of the modelling work were to:
Model views and entity-attribute definitions were presented in the CIRCE Project Report1.
A "bottom-up" view of data
Initial work in this area focused on Data Flow Diagrams. However, it rapidly became apparent that this approach was not going to meet the information needs of the project, which was to correlate the shared need for information amongst a wide range of health and social care workers in a wide range of settings. A Complex Object Grid (COG) was developed as an alternative method of modelling data2. This analysed the clinical and administrative data items on the basis of each individual group of health and social care worker's requirement to generate, use and need to know data. It was presented as a series of spreadsheets that were based on sections of the record. The main benefits of the COG approach were that it provided:
Further IS development work on CIRCE
The CIRCE Project Consultants at the School of Information Systems produced a high-level system architecture supporting the concept of a Federated Database that permits data sharing within the context of a defined set of agreements between "information trading partners". This architecture also defines a framework which could support a more fully developed Electronic Patient Record (EPR)3.
Prototyping development on CIRCE
Prior to the CIRCE Project, prototype development work had taken place in a hierarchically-ordered database based on B-Trieve (Protos Patient Care System). A series of applications were developed for discharge from hospital, social services, and a district nurse led leg ulcer clinic using Version 2 of the Read Codes as the clinical dictionary. Development in general practice was limited to a simple form for referral to the district nurse. It was intended to mount trials of these applications in a live environment. Unfortunately local changes in community health services management structure prevented this happening. This led to a switch in project resources into the development of a prototype demonstrator in a relational database environment (Microsoft Access). This new prototype was used successfully in a live trial at the local community hospital. It was also part of a laboratory demonstrator at the University of East Anglia that demonstrated the principle of simple message passing between three different databases.
The Problem- and Episode-based Recording System for Electronically Updating Summaries (PERSEUS) Project
After the Norfolk and Norwich NHS Hospital Trust assumed formal ownership of the Project, further development was required in order to have access to the Version 3 Read Codes in the prototyping environment. The CIRCE IT Project Assistant developed two data bases: a Read Code database in which to store the Codes, and a Browser that provided access to the Codes via term keys. There were significant problems with search times on the initial browser and, on advice from Computer Aided Medical Systems (CAMS) Ltd., the data was re-ordered to improve system performance. Even with this modification the browser will not run sufficiently fast to cope with the rigours of input in a ward environment. Datasets of terms and codes are therefore prepared in separate tables and accessed directly from the clinical application as "look up" tables.
Hospital prototyping trials
The following trials of prototyping are intended during the PERSEUS Project
General practice summary generation
Work in general practice is centred around the process of notes summarisation at a general medical practice in Dereham, Norfolk. Initial work has concentrated on developing a simple application to emulate the current practice word-processed template. The process in this database is based on the traditional written summary that is constructed de novo and then updated periodically. The next stage of the process will be to build a prototype application that creates a record of all data received in the practice from external sources over a period of time and also contains a record of encounter-based problems over that same period. Summarising this record will provide a more accurate picture of the IS/IT scenario following the introduction of NHS Clinical Messages. Given a suitable clinical record model generation of these messages should not produce significant problems for the sender of the message. The reception and subsequent management of the data held in the messages by the recipient is a different matter. There is the potential for massive duplication, and reduplication, of data within the NHS. The selection of data items for inputting into the GP database (from NHS Hospital Discharge Messages) and for inputting into the hospital database (from GP Referral Messages) will require careful consideration. It is intended to create a series of fifty records to study the relationship between episode based problems and diagnoses and problem lists over time. This study may provide some data on these important areas of future development.
Wider national IM&T initiatives
The project team has also been involved in a number of national coding and EPR developments including the Core Headings for the Personal Health Record (Department of Health). Currently it is undertaking an evaluation project for the Information Management Group on the use of Read Codes in the hospital discharge process. This latter project has permitted further development of a local coding interface for the BCSR Workstation and it is intended to incorporate these developments into the next GP Summary prototype.
Conclusion and summary
The audit work described in this paper was part of a continuing IS development project relating to interdisciplinary and cross-boundary information management. The audit work was principally concerned with manual discharge and referral records. It provided evidence that structuring records can improve the quality of discharge and referral documentation. Computerising these records will produce further enhancements to the quality of health records.
It is important for any IS development to start from a sound base. This project has selected clinical communication as the starting point for an analysis of EPR information requirements. The project has worked, and will continue to work, within the framework of the NHS Information Management and Technology Strategy. The IS methods are based on evolutionary systems design which utilises rapid prototyping as one of the techniques of development. These prototypes require to be tested in a live clinical environment. Trials are currently in progress at the Norfolk and Norwich Hospital Healthcare NHS Trust.
References
1 Keele G. The Computerised Integrated Record for the Care of the Elderly Project (CIRCE) - Final Report to the NHS Information Management Group. January 1995
2 Keele LR, Gilbert D, Smith DJ. Development and Impact of a Model for Shared Health and Social Care Information. In Proceedings of the Sheffield Health Information Management and Research Symposium, 1995:166-181
3 Smith DJ, Dearnley PA. The CIRCE Database Federation: data sharing for health and social services. In Richards B. (ed.) Proceedings of Healthcare Computing '95 Conference. BJHC Ltd., Weybridge, 1995:540-549