AAU logo

V-CHI årsmøde 2001

V-CHI årsmøde 2001

Torsdag den 1. november 2001 på Aalborg Universitet.

Tema: rundt omkring den elektroniske patientjournal


Fra Masteruddannelsen i Sundhedsinformatik på Aalborg Universitet blev nedenstående projekter præsenteret:

1) Klassifikation & Kommunikation – en diskussion om en fælles sundhedsfaglig klassifikation
forfattere: Lone Asp, Annelise Kjølseth Kent, Mona Brøllund Tewes

Se også projektet i pdf-format.


The quality of care depends largely on how different groups of health care professionals can communicate and co-operate based on the documentation
in Electronic Health Care Records of patient conditions.

The purpose of this project is to examine the problems associated with a multiprofession-nal approach in using the official classification system in the
health care sector in Den-mark. As nurses experienced that the only possible classification of patient conditions, ICD-10, is not suited to document
nursing diagnosis.

A number of authors have proposed criteria for evaluating standardized coding and clas-sification systems to support clinical practice. We have
narrowed these down to seven common criteria for a good clinical classification for documentation in Electronic Health Care Records and tried the
classification of diseases, i.e. ICD-10.

As this classification failed most of the criteria we came to the conclusion that the struc-ture of this classification did not meet the needs of classification
and coding of clinical in-formation. Based on classification theories we recommended a more flexible facetted structure for a classification scheme for a
multiprofessionnal classification system.

We nevertheless ended up sketching a model for a ”translator” to meet the actual clinical situation where health care professionals as doctors and
nurses have their own standard-ized languages and a common understanding of classification schemes is not apparent.


2) E-learnings rolle i en undervisningsstrategi ved implementering af den elektroniske patientjournal
forfattere: Irving Hempel Bengtson, Pia Wichmann Madsen, Nicholas Faddy


The coming widespread implementation of the computer based patient record in Denmark will be accompanied by the requirement to train a large
number of health care workers in its use. The most commonly used type of IT-course is based on the ”hands-on” principle – that is, a classroom-based
session involving a lecture followed by practical exercises. This type of course has been found to be inefficient both in terms of learning value and the
use of resources. Furthermore, the particular circumstances of the health-care sector, such as shift-work, many part-time workers and not least the
general lack of IT skills, creates a need for flexibility in the planning of teaching which is difficult to meet.

We wish therefore to investigate the possibility of using other teaching methods that may meet these needs, and at the same time offer better learning as
well as more efficient use of resources. One such method, which is becoming increasingly popular in the private sector, is e-learning, which we examine
with a view to its use in combination with classroom methods.

A series of qualitative interviews with nursing staff from a hospital department where an electronic patient record has recently been introduced has
confirmed our original problem definition. With this in mind, we have examined the respective advantages and disadvantages of e-learning and
traditional IT-teaching with respect to the electronic patient record, and suggest ways in which such a teaching programme may be planned.


3) Beslutningsstøtte med Bayesiansk netværk: en model til tolkning af laboratoiresvar
forfattere: Ram Benny Christian Dessau, Maj-Britt Nordahl Andersen

English summary

Decision support using a Bayesian network. A model for interpretation of diagnostic tests.
A case of serology for Borrelia infections (Lyme disease).

Many laboratory tests, do not have a 100% sensitivity or specificity, and false negative and positive results may occur. It well known that some
serologic tests are far from perfect. The size of the problem depends on the prevalence of disease in the population under investigation. There is no
tradition for measuring the number of false results occuring during routine testing, even this is important for the diagnostic decisions.

An important aspect is the interpretation of probability as subjective, this is the probability of a certain state or diagnosis in the individual patient, not the
mean of a group. Bayesian statitics may be used to calculate predictive values for the results of a diagnostic test. This method of computation, may be
extended to handle many variables and computer-based graphical tools are available for complex modelling.

Problem definition
Is it possible to construct a statistical model to calculate predictive values for test results tailored to the individual patient ? The model should support
the natural course of the clinical diagnostic process.

Building of the model
A probalistic model for the diagnosis of Borrelia infections was built using HUGIN (Hugin expert A/S, www.hugin.dk). Preliminary data for the model
was collected from the litterature, an interview with clinicians concerning 47 patients both positive and negative on ELISA for Borrelia antibodies and
the laboratory database on approximately 3000 specimens and subjective estimation of unknown variables. Seven clinical casedefintions (European
concerted action on lyme borreliosis), the temporal progression of infection in four stages, are modelled together with relevant elements of patient
history and background data (season, age, sex). During the development an object-oriented description of the model was implemented.

The model has been evaluated using constructed ”typical” clinical cases. The logical and relative consistency of the model makes sense, however the
magnitude of changes in probability is not yet well adjusted, as data are missing. A larger clinical investigation has been planned for data acquisition. An
important point is that this investigation examines all the patients in whom Borrelia test are ordered, as patients with negative results and low prior
probability of Lyme disease are important for calculating the predictive values. The model is not yet ripe for clinical evaluation.

The model should be used in conjunction with electronic ordering of tests, to help the clinician choose the appropriate tests by calculating predictive
values based on the clinical information for the particular patient. An automatic and interactive connection between the decision support module and the
health care record should be established. This model makes proposals for minimum clinincal datasets which should accompany the ordering of
diagnostic test. It is the combination of clinical data and laboratory data that gives a new dimension to interpretation of the diagnostic test. Also the data
for the probalistic model should be kept current as the epidemiology af Tick bites and the exposed population will change as a function of time.

The main perspective of this project is that laboratory results, with a categorical interpretation, should be accompanied with predictive values for the
individual patient based on the current local epidemiology. A probabilistic model is proposed. It is a tool to support interpretation of the diagnostic
value of a particular laboratory test, as an improved basis for the clinical diagnosis. The implementation of the decision support system is dependent on
a realtime combination/integration of the decision model with laboratory data and clinical information when ordering the test.


4) Proaktiv teknologivurdering – Brugergrænseflader
forfattere: René Jensen, Susanne Knudsen, Dorthe Lassen, Dorthe Wøldike, Marianne Arnfjord


Information technology, in particular the field of utilization, is characterised by a high level of complexity and continous change. Applied models for
technology-assessment should reflect and express methods which are based on a non-linear philosophy. In the field of information technology the use
of an iterative point of view seems to be a core aspect, as well as the combination of several different methods in the process of assessment for the
purpose of gaining a high quality level of analyses

The scope of this project is to make a proactive technology assessment of an already existing technology: a graphical user interface. At the same time
we wish to evaluate the possibility of making the technology assessment with a proactive approach.

The applied metods seem to be an appropriate way of starting a process. A major result of the analyses emphasises the need to view the chosen
methods as ’starting-engines’ in a longer and more complex process of the innovation and implementation of information technology.
Such a strategy supports an ideological basis of participative design. Participative design and the importance of an appropriate organization are other
important results of the analyses – especially for the purpose of making the process iterative and of creating motivation throughout the organization.

To indentify the history and premises of a large scale innovation- and implementation process is crucial, due to the possibilities of controlling the
technology assessment proces in itself - from the starting point, towards a defined goal.