Article Text



Statistics from


A.D. Cowley, S.V. Beath, C. Cummins2, D. Porter, I. van Mourik, P.J. McKiernan, D.A. Kelly.

The Liver Unit and 2 Statistics Dept, Birmingham Childen’s Hospital, Steelhouse Lane, Birmingham, B4 6NH

Aims: To develop an audit tool, which measures disease severity and clinical dependency in children with liver disease.

Methods: The initial score was based on known prognostic variables (J Malatack, et al) used in acute and chronic liver disease. Between 1997 and 2001 the score was modified four times and reduced to comprise 10 key variables as follows: Aspartate transaminase, prothrombin time, albumin, bilirubin, ascites, nutrition, organ dysfunction, blood product support, sepsis and IV access. The first five parameters are specific to liver disease and the last five are generic dependency parameters. The score was validated by examining 1) face validity, 2) construct validity, 3) internal consistency (Cronbach’s Alpha). Inter-observer differences were evaluated by the methods of Bland and Altman.

Results: The final 10 variable score was tested on 132 (all children admitted to the ward over two 3 month periods). The score was valid and internally consistent. The Cronbach’s Alpha score for the total score was 0.72. It has good construct validity and correlated with the length of stay (Spearman’s correlation, 0.731 significant at 0.01 level). There was also a significant correlation between the liver and dependency score (P<0.01). The score was easy to collect, reproducible and the inter-observer error was acceptable.

Conclusion: The PHD Score is valid and internally consistent and has the capacity to measure disease severity and dependency prospectively. It is now being recorded on all summaries in order to facilitate departmental audit.

Acknowledgement: We are grateful to D. Sandhu, C. Lloyd and C. Weijers for their help.


P. Ramnarayan, A. Tomlinson, G. Roberts, M. Coren, V. Nanduri1, T. Sajjanhar2, J. Britto.

Imperial College London at St Mary’s, 1 Watford General Hospital and 2 Lewisham Hospital

Background: ISABEL ( is an internet-delivered decision-support system that provides a novel differential diagnosis tool. In response to clinical features from a patient, it produces a list of 10–12 diagnoses for the clinician’s consideration, by searching 4 standard paediatric textbooks.

Aims: To assess changes in clinicians’ decision making quality after their consultation with ISABEL, for a representative set of case simulations based on real patients.

Materials and methods: 24 case histories (12 different specialities and 3 levels of difficulty) were used. Clinicians of different grades (Consultants, SpRs, SHOs and medical students) were recruited by invitation from the ISABEL UK user database. On a special trial website, subjects recorded their clinical decisions (differential diagnosis, investigations and management plan) before and after consultation with ISABEL. A panel of two general consultants provided “gold standard” decisions against which the subjects’ decisions were measured, using a previously validated scoring system.

Results: 76 subjects (18 consultants, 24 SpRs, 19 SHOs and 15 students) completed 751 case episodes. Improvement in decision making was maximal for students and inexperienced doctors, and moderate for consultants. Median extra time taken for ISABEL consultation was 1 min (IQR 30 sec to 2 min). See table.

Abstract G126

Conclusions: ISABEL reminded doctors of a significant diagnosis not previously considered in 1/8, and significant tests in 1/10 cases in the simulated environment. Extrapolating these results to real life suggests a significant improvement in clinical decision making in acute paediatrics.


R.I.J. Paget, P. Ramnarayan, M. Coren, I. Maconochie, A. Windrow, J. Bucknall, M. Tomlinson, H. Fisher, D. Broomfield, J. Britto.

St Mary’s Hospital NHS Trust, London; The Homerton Hospital NHS Trust, London; Kingston Hospital NHS Trust, London

Background: The Internet provides a vast new resource for clinicians. Its use is greatly enhanced when readily available at the bedside, in the clinical workflow.

Aims: (1) Survey the current status of Internet access in UK Paediatric units. (2) Study the advantages of using Internet connected pocket sized computers, over Desktop PCs in 3 NHS hospitals.

Methods: (1) In conjunction with the DoH, questionnaires were sent to a sample of Paediatricians across the UK, who were asked about their computer resources, access, location and skills. (2) Desktop PC stage: The usage of a Paediatric Clinical Decision Support System ( by 46 junior Paediatricians in 3 hospitals on their desktop PCs was monitored for 7 weeks. Pocket PC stage: Windows pocket sized computers with a wireless, aerial link to the Internet for 7 weeks (ongoing).

Results: (1) Hospital Paediatricians returned 48 questionnaires. 48% were Consultants. 54% work in District General Hospitals. 98% use computers > once a day and 79% described their computer skills as good. 8.3% had no Internet access at work, and 54% share these computers with >6 other staff, >20 staff in 30%. In only 25% were the computers in the same room or ward as their patients, and in another building in 34%. Over 75% said they would use a Paediatric Internet resource (such as ISABEL) if they had more opportunity to do so. (2) Desktop PC stage: In a 7 week period subjects logged on 98 times but described lack of computer access as their greatest limiting step. Pocket PC stage: Initial results of the ongoing pocket computer experiences have been very positive.

Conclusions: Our study suggests ease of Internet access in the UK units surveyed is well below that recommended by the DoH. Lack of space/mobility limits introduction of near patient desktop PCs. A computer in every clinician’s pocket, permanently ‘online’, will bring the Internet’s corpus of medical knowledge and decision support systems to the bedside and into the workflow.


A. Greenough.

Caldicott Guardian, King’s College Hospital, London, UK

Background: Patients have the right to expect confidentiality. The principles essential to those handling patient information in clinical practise, teaching or research are outlined in the Caldicott Report and the Data Protection Act.

Objective: To determine if supervisors of trainees and teachers of undergraduate and postgraduate students were aware of the principles of the Data Protection Act or the Caldicott Report, particularly with regard to the electronic storage of patient data.

Methods: Consultants at a teaching hospital were surveyed and asked to grade their level of knowledge (1 = little/none, 2 = adequate operational, 3 = detailed) of the Data Protection Act and the Caldicott Report. They were also asked whether they stored data electronically and for what purposes and if they had training needs regarding use of patient identifiable information.

Results: 74 consultants completed the survey; 96% were aware of the Data Protection Act (mean knowledge level 1.7) and 58% of the Caldicott Report (mean knowledge level 1.2). Eighty-eight percent held some form of patient information electronically, those consultants most commonly indicated that they had training needs (p = 0.031). Fifty-six percent stored patient identifiable information on computer for research purposes and were more likely to also store patient data off site (p = 0.0006).

Conclusion: Caldicott guardians of all NHS Trusts should ensure there is teaching on confidentiality in induction programmes for new staff and in-house training for existing staff.


G.D. Baxter1, K. Tan2, S. Newell2, P.R.F. Dear2, A. Monk1.

1 University of York; 2 St James’ University Hospital, Leeds

An expert system (ES) is being developed to empower the decision making (DM) of front-line carers (FLCs)—SHOs and nurses—treating mechanically ventilated babies with RDS in a Neonatal Intensive Care Unit (NICU). The ES detects blood gas and ventilation anomalies, and suggests remedial changes to the ventilator settings.

Staff acceptance of the ES will be affected by how well it integrates with their normal work. After carrying out a rich picture analysis of work on the NICU identifying staff roles and responsibilities (reported elsewhere), the DM processes involved in changing ventilator settings have been analysed using a semi-structured interview technique, the Critical Decision Method (CDM).

Interviews with four FLCs and four experts (registrars and consultants) were recorded. The CDM was explained before they were asked to recall an incident where they changed the ventilator settings. A sequential time line of the incident was developed and critical decisions identified and verified. Staff were asked probe questions about available options, information requirements, goals, and time pressures for each decision point.

The data were used to create sensory cue inventories and situational assessment records for each incident. Experts, who are called in when the FLCs cannot solve the problem, use more cues than FLCs. This can be explained by their need to review the situation (using verbal and visual cues) before looking for further information to diagnose a particular case. Two experts reported changing their diagnosis in light of new information. Experts also reported being prepared to make bigger changes to the ventilator settings than FLCs.

There are two main implications for the DM of the FLCs in the NICU. If the FLCs are to learn why the ES made a particular decision, they must be able to explain the reasoning behind each decision. Furthermore, since FLCs currently refer difficult cases to the experts, the FLCs will need to know the precise limitations of the ES so they can decide when to handle the case (using the ES) and when to refer it to the experts.


K.G. Ravikumar, M.J. Pocha.

Department of Paediatrics, Bedford hospital, Bedford

Background: It is common practice for hand-written discharge summaries to be sent to GPs about a child’s admission to hospital, diagnosis, investigations, discharge medication and follow up arrangements. The information in these letters is often faint, illegible and incomplete. Dictating the summaries for subsequent typing is time-consuming. Some existing computer programs were tried and found not to be flexible or user-friendly.

Aim: To assess the use of a tailor-made, user-friendly computer program to produce quick and legible discharge summaries to GPs.

Methods: A ‘macro’ was created in Microsoft Word, using Visual Basic for Applications (VBA). The program accesses the main database (Patient Information Management System) to get patient related data. A selection of common paediatric referral pathways, symptoms, diagnoses, treatment, investigations and discharge medication is presented in listboxes according the system (respiratory, cardiac, etc.), which users can choose. Most of the common paediatric problems can be clicked and added to create a summary but there is also the ability to add new items and free text, to give maximum flexibility, speed and ease of use. The program was introduced in August 2000 and refined over time. 85 consecutive patients admitted to the ward in November 2002 were studied and questionnaires sent to SHOs and GPs.

Results: Most new SHOs found it quick and easy to use after a brief period of training. 70% of children had discharge summaries done at the time of discharge and the rest had hand-written summaries, to be followed up by a printed summary. Average ratings given by GPs on the quality of summaries were 8, 9, 9, 9, 8 (out of 10) for promptness, clarity of patient information, clarity of clinical information, readability and usefulness respectively. The software program will be demonstrated and discussed.

Conclusions: Printed discharge summaries are the best way to improve clarity, legibility and completeness of information provided at discharge. A computer program which is easy to use, fast and flexible is essential for junior doctors to complete summaries at the time of discharge.


A. Fox1, P. Davies2.

1 St Mary’s NHS Trust, Paddington, London; 2 Bedford Hospital NHS Trust, Bedford

Background: The Internet is the future of medical knowledge sharing. This is reflected in the development of subscriber-access online publishing by the majority of reputable medical journals. However, a new breed of journals is emerging which aims at an international audience, and is free to access.

Discussion: The founding ethos of the Internet is the free exchange of information. The established medical journals use of subscriber only access is justified by the costs of producing high quality peer reviewed content. This model limits the proliferation of information to the privileged few, and globally, excludes the majority of those who would benefit from the information they contain.

The development of online medical journals releases the publisher from many of the costs traditionally incurred by the production of a paper journal. They are free and easy to view, wherever the user may be in the world. This promotes the ideals of inclusivity and accessibility that are espoused by the Internet.

Rigorous peer review need not be sacrificed by well run free journals. The greatest barrier to the growth of the stature of these journals is the limited volume of high quality submissions. This stems from a tendency for authors to seek peer recognition for work by submission to a recognised and established ‘hierarchy’ of medical journals. Free to view Internet journals have a far greater potential true impact factor that any restricted journal. We review the current status and proliferation of all types (traditional and new) of journals available on the internet, and discuss the future status thereof. Paediatricians should be aware of the ease of use of these journals and the shift towards sharing medical knowledge to a truly international audience.

This discussion is drawn from the authors experience as editors of the “Internet Journal of Pediatrics and Neonatology”.


D.C. Brown, E. Dhouieb.

Royal Hospital for Sick Children, Edinburgh, UK

Background: Hospital staff were surveyed to assess which had been the most effective means to develop awareness and use of a new intranet training site. They identified the most user-friendly aspects of the site and those pages most relevant to their training needs.

Description: During redesign of the hospital’s acute medical receiving unit, a multidisciplinary workshop identified staff training as essential to facilitate introduction of new care pathways. A training microsite (55 pages, file size 3.6 MB) was launched, using Microsoft FrontPage 2000, on the trust’s children’s services intranet.

Results: 36 doctors, 38 nurses and 40 staff from therapies, administrative, laboratory and clerical groups were surveyed. 69 (60.5%) responded. 23 (33.3%) responders were unaware of the site. Staff had usually been made aware of the site by colleagues, email circulars and presentation at hospital education sessions. The appearance of the site, ease of navigation and relevance to personal training were rated 7.4±0.4, 7.6±0.4 and 6.3±0.5 (mean±SEM) out of 10. The three most popular topics on the site were Picture of the Month, Review Articles and Lectures & Meetings. Calm colours and the absence of busy graphics were preferred. Undergraduate students felt excluded by the index pages’ links to (graduate) staff groups. Classification of pages by disease process, clinical service or department was preferred to that by staff group. External access to the site from home or other hospitals was requested. Publication of the survey results increased staff awareness of the site further, and led to offers from staff to compose their own pages. The site was redesigned to take account of the survey results.

Conclusions: There was a positive reaction to the creation of a staff training microsite. Identification of favoured features encouraged staff to expand the site themselves. External links from other health providers and to the Internet are being considered.


R.D. Palmer, A. Maw.

Department of Community Paediatrics, South Cambridgeshire Primary Care Trust, Ida Darwin Hospital, Cambridge

Introduction: The NHS Plan and the Kennedy Report (Bristol Children’s Heart Unit Public Inquiry) have both referred to the policy of copying letters between physicians to the patient concerned. The Department of Health (DoH) maintains that this is a patient’s right and should become standard practice, so a Working Group has been set up in order to implement this policy by April 2004.

Aims: Whilst patients and parents already have access to their medical records, few seize the opportunity. We wished to explore paediatrician’s current practice, views on this policy and their perceptions as to whether this would impact the service they provided.

Methods: A postal questionnaire requesting information on current written communication practice along with opinion on the DoH plans was conducted over a 4 week period. The questionnaires were circulated to all Consultant, SpR and Staff Grade paediatricians in South Cambridgeshire, covering secondary and tertiary services across two health care trusts.

Results: The response rate of 83% (48/58) was equivalent across grades. Currently 17% do not copy letters to parents, with 54% occasionally, 23% frequently and 6% always doing so. However, 22% copy letters by writing a completely separate letter to the GP and parent, perceiving language and style a problem, something more a half of all respondents were concerned about if this does become routine. In terms of the DoH plans 79% do not believe it should become compulsory, although 65% would be more than happy for their most recent clinic letters to be copied unedited to the parents.

Conclusions: Copying letters to parents is likely to become mandatory, so training and resource issues need addressing now. Further, parental desires should be sought so that we don’t expect them to fulfil the professional caring role just because they are perceived better informed, in writing. Specific issues such as child protection, social or psychological concerns, communicating uncertain diagnoses or conflict needs consideration to avoid undermining the GPs professional contribution and continuity of care.


K. Tan, S.J. Newell, P.R.F. Dear.

Neonatal Unit, St James’s University Hospital, Leeds LS9 7TF

Aims: To assess the strength of evidence for the effectiveness of clinical decision support systems in the care of the newborn infant by conducting a systematic review of the published literature.

Material and Methods: A protocol and a search strategy were developed (modelled after the Cochrane Collaboration), in October 2001 and repeated in August 2002, to find clinical decision support systems (CDSS) that were developed for neonatal care. Only studies that have been evaluated (or tested) in the clinical environment were included. Studies without a control arm were not excluded. The following electronic databases; Medline (1966-August 2002), Embase (1980–2002), CINAHL (1982 to August 2002), HealthStar (1973 to August 2002), AMED (1985 to August 2002) were searched as were proceedings of international meetings. Level of evidence was graded after the Oxford CEBM criteria.

Results: The above search strategy produced 447 citations initially which was reduced to a total of 67 that met the inclusion criteria. Regarding the papers selected, 13 systems related to prescribing category; 11 were for prognosis; 20 for ventilator management and automated monitoring; 2 for management of the infant on ECMO, 1 for neonatal transport, 2 for management of congenital heart disease, 10 for aiding diagnosis; and 8 miscellaneous. Three of the systems were evaluated by a study where either the patients or doctors were allocated to a control group. The majority of the systems showed benefit for patient care.

Conclusion: This review of the published literature showed that computerised decision support systems have been developed for many areas of neonatal intensive care. Such systems may confer benefit. However, methods used for evaluation of clinical effectiveness have not been sufficiently robust.


K. Tan1, G. Baxter2, K.G. Brownlee1, S.J. Newell1, P.R.F. Dear1, S. Smye1.

1 St. James’s University Hospital, Leeds LS9 7TF; 2 University of York, Department of Psychology

Aims: Mechanical ventilation of newborn infants is adjusted according to the level of blood gases and this process is governed by rules. Expert systems are therefore feasible in this situation. Fuzzy logic based expert systems are widely used to provide rapid solutions in complex situations based upon user-defined rules. Our aim is to develop a computerised decision support system for ventilation of the newborn infant.

Methods: A multi-stage fuzzy logic expert system (FLORENCE) was developed using MATLAB 5.3 (Mathworks, Natick, MA). A set of 20 scenarios featuring different oxygen/carbon dioxide abnormalities under different ventilatory conditions was created. The fuzzy logic expert system was tested on these scenarios in the Simulink™ environment. The answers derived were then graded by a panel of neonatologists and respiratory paediatrician, who had evaluated the scenarios beforehand.

Results: The fuzzy logic expert system was judged to give acceptable solutions to 17 of the 20 scenarios. In the 3 where there was disagreement, it was agreed that the solution from the system would lead to partial correction of the blood gases. None of the solutions in the 3 cases were potentially unsafe. There was precise agreement with the solution provided by the expert system in 6 of the 17 cases; the remaining 11 cases gave qualitatively correct and clinically appropriate answers.

Conclusion: A decision support system for ventilation of the newborn infant using fuzzy logic is feasible. This pilot study is being followed up by a larger scale assessment using the DELPHI technique.

View Abstract

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.