Learning support for the consultation: informationsupport and decision support should be placed in aneducational framework
Paul Robinson,1 Ian Purves1 & Rob Wilson2
Background Advances in information technology mean
that it is now possible to provide contextually relevant,
evidence-based information during the course of the
consultation. As a consequence, the practitioner has to
consider the new information (from the computer) in
the situation of the present consultation and in the light
of his or her own experience. This task has to be carried
out in a short time, in the presence of the patient.
Method Drawing on experience of the development of
one decision support system, this paper places that task
for the practitioner in an educational framework. We
begin by reviewing theories of professional experience
and knowledge and go on to look at schema theory and
the role of cognitive dissonance and reflection in
learning.
Conclusion This paper considers the provision of real
time decision support in the light of learning and the
experienced practitioner. We conclude that framing the
implementation of decision support in this way pro-
vides useful insights. The key process is learning by the
practitioner, in the course of the consultation. This
process should be supported by decision support and
information support software. There are implications
here for the design of such software, and also for the
way in which practitioners are trained to use it.
Keywords education, medical, continuing ⁄ *methods;
physicians, family ⁄ *education; decision making; profes-
sional practice ⁄ *standards; *learning; Great Britain.
Medical Education 2003;37:429–433
Introduction
Primary care in the UK is mostly computerised (in 2000
99% of GPs used computers for repeat prescribing and
85% used them, at least partially, for making medical
records)1 and has provided a setting for the development
of computerised decision support systems. It is now
possible to provide contextually relevant, evidence-based
information during the course of the consultation. As a
consequence, the practitioner has to consider the new
information (from the computer) in the situation of the
present consultation and in the light of his or her own
experience. This task has to be performed in a short time,
in the presence of the patient. The key process is learning
on the part of the practitioner, in the course of the
consultation. This paper reviews theories of experience,
knowledge and learning. It goes on to consider how
learning by the practitioner, during the course of the
consultation, might be supported.
In their review, Delaney et al.2 describe the potential
for computerised decision support to deliver improved
quality, but conclude that this potential has not been
realised. Although we would debate Delaney et al.’s
view, especially in the context of PRODIGY,3 we believe
there is great potential for computerised decision
support. The systems so far developed fail to meet the
needs of the practitioner and the patient in the primary
care consultation: a �sophisticated understanding of the
process of the consultation is required�.2 We have in the
past described such a process for the consultation4 and
believe firmly that the �system� is much wider than the
computer, its software and its knowledge bases: it
involves both the clinician and the community in which
he or she lives. In this paper, we suggest that what is
needed is not decision support, nor information sup-
port, but learning support.
The ideas that are developed in this paper arose from
an iterative development of the PRODIGY project and
from research into the use of knowledge during the GP
1Sowerby Centre for Health Informatics, University of Newcastle, UK2University of Newcastle Business School ⁄Centre for Software
Reliability ⁄Centre for Urban and Regional Development, University
of Newcastle, UK
Correspondence: Dr P J Robinson, Wayside, Pickering Road East,
Snainton, Scarborough YO13 9AF, UK. Tel: 00 44 191 243 6000;
Fax: 00 44 191 243 6101; E-mail: [email protected]
Learning support
� Blackwell Publishing Ltd MEDICAL EDUCATION 2003;37:429–433 429
consultation.3,5 The aims of the first two phases of the
PRODIGY system were to deliver prescribing therapy and
condition management advice (�clinical recommenda-
tions�) to the general practice clinician at the point of
care, in this case the consulting room. The general
practitioner (GP) utilising the system during Phase 1
was presented with therapy options usually consisting
of three drug preparations, but for some conditions the
option of a patient information leaflet was also available
as an alternative. Extensive evaluation, using a pre-post
intervention trial model, was carried out.6 This inclu-
ded focus groups, video-recorded consultations, user
questionnaires and computer generated interaction log
files of usage. This led to both improvements (including
the insertion of �scenarios�) and greater understanding
of the participating GPs’ use (and non-use) of the
system, leading to a Phase 2 trial system and thus
completing one iterative development cycle. The same
model was then used to evaluate the Phase 2 system
before culminating in a nationally available version
(Release 1). PRODIGY is a mixed cognitive and rule-
based system, which is founded on guidelines that
advise on therapeutic actions such as prescriptions,
advice leaflets, referral and investigation.
The research into knowledge use during the consul-
tation5 found, in nearly all consultations studied,
instances of cognitive dissonance between action that
the GP knew to be recommended by published work
and action that was based on his or her own experience.
Practitioners nearly always chose to follow the latter.
The difference between guideline recommendations
and what the practitioner in fact decides to do is
analogous to the research–practice gap.7 We work
from the premise that, in order to make good use of
on-line decision support, the practitioner has to learn
in two areas: how to incorporate into the consultation
information derived from the computer, and how to
adapt clinical practice to the guideline recommenda-
tions.
The GP’s experience
Even the newly appointed GP Principal has consider-
able experience of and expertise in medicine, consulting
with patients and working in primary care. The
development of expertise in medical practitioners and
the role of experience in learning have been well
described.8–10 Qualitative analysis of knowledge use
by GPs during the consultation5 shows that, during a
consultation, the GP’s conceptualisations interact with
and integrate knowledge derived from four different
sources: the literature, colleagues, the patient and his or
her family, and the doctor’s own memory and experi-
ence (see Fig. 1). These sources (excluding colleagues)
correspond to Popper’s World 3, World 1 and World 2,
respectively.11 That is, the written record and the
literature are objective structures produced by people;
the symptoms and signs of patients are, for the GP,
material things; and memory and experience comprise
the subjective world of the mind.
Although this model is cognitively derived, it is
consistent with social learning theory.12 Learning is
situated within the consultation, and the participants in
the consultation constitute a community of practice.13
Communities of practice are forums in which joint
participation leads to shared understanding and defini-
tions of meaning. The doctor is also a member of other
communities of practice, namely the participants in
other consultations, the primary care team and the
medical profession.
Key learning points
Dissonance is a prerequisite to learning.
Dissonance is emotional and frightens people off.
This provokes a need to help GPs develop schema
to cope with dissonance.
Learning is situated in the consultation.
Learning support needs:
• to tell a story (intuitive) based at population
level and adapted to the individual patient (if it can
be);
• to support just-in-time learning, and
• to induce post-decision dissonance and support
constructive learning (i.e. through action support;
reflection support after consultation).
Figure 1 The GP’s experience; literature refers to the written
medical record as well as to papers and textbooks.
Learning support for the consultation • P Robinson et al.430
� Blackwell Publishing Ltd MEDICAL EDUCATION 2003;37:429–433
The implication of Fig. 1 is that the GP’s current
experience should be informed by all four sources.
Unless written material or colleagues’ advice or infor-
mation about the patient is directly sought during the
consultation, it is represented in the consultation via
the GP’s memory. The literature is considered part of
the GP’s experience, rather than separate and different
from it. Alternatively, it is possible for the GP to relate
only to his or her own memory and the patient. This is a
mechanism by which rigid habits can develop, to the
exclusion of research or audit findings. The social
context of the GP consultation may encourage this.14
Experience and expertise constitute a conundrum. On
the one hand, they are a practitioner’s greatest assets;
indeed, in Kant’s view experience is all that any individ-
ual can have. On the other hand, �experience� may be
synonymous with resistance to change.15 Straddling this
divide is the view that teaching is most effective when the
learner’s previous experience is taken into account.16
As Dowie7 describes, research-derived knowledge is
fundamentally different in character from practitioner-
based knowledge. The experienced practitioner has a
knowledge structure that is based on illness scripts or
case memories.8 These are remembered and related as
stories, and there is a new appreciation of the import-
ance of narrative to medical practitioners.17 Stories are
built up from schemata, which we describe below.
Schema theory
Any perception of the world is interpreted through a
person’s schemata, or organised knowledge of the
world. These schemata are the basis of comprehension,
learning and remembering. Ideas sit in the �stories� in
which the human mind embeds its sensory percep-
tions.18 They can be viewed as structured expectations
about people, situations and events.
There are differing types of schema:
1 Person schema, in which we hold our perception of the
characteristics (skills, competencies and values) of an
individual. �Self � is a specialised person schemata.
2 Event schemata, or cognitive scripts, that tell us how
to behave in differing situations.
3 Role schemata, which are third-person event sche-
mata or models that we can use to predict how others
might behave in certain situations. We can apply
these to simulate �event schemata� and this can be a
route to the development of personal event schemata.
These schemata are developed either directly from
personally refined (act–observe–review) events or indi-
rectly through others’ stories. Stories are built around
actions taken by �actors� and ⁄or �things� within an event.
Stories also have plots which link events backwards in
time and �predict� potential future events. Of course,
stories can be narrated using many differing forms of
discourse, whether it be thought, spoken, written or
acted using a variety of media.19 Initially, schemata are
simple but they become more complex, interwoven and
�specialised� over time. Schemata are in effect narratives
with the discourse as the distributed representation in
the brain.20,21
Everything we perceive is pattern-mapped into our
many schemata and, as we become more experienced,
we develop highly specific schemata that enable inter-
pretation – when we are less experienced we use less
refined schemata.8 This human mechanism is highly
effective and, despite our amazing developments in the
computing field, is not reproducible, on a usable scale,
even at an embryonic level.22 However, the �human
mechanism� does have drawbacks as everything we
perceive is filtered by our schemata to the extent that, if
we have well-refined and heavily weighted schemata
that are relevant, we can �throw away� perceptions that
do not �fit� our experience.
Of course this implies that schemata can be fairly
rigid: nothing is further from the truth – they are
dynamic. There is a spectrum between no experience
(i.e. no relevant schema) and expert (i.e. highly
specialised schema). If we have no schema we develop
new ones, we find a schema that partly matches, and we
adapt it and create a further specialisation. Problems
occur if we have inappropriate rigid attitudes (self
schema), or perceptions or advice that conflict with
highly specialised schemata. In this situation there is an
emotional discomforting response of cognitive disson-
ance. How we deal with this dissonance determines
whether we learn.
Cognitive dissonance
Cognitive dissonance is the perception by a subject of a
difference, of variable intensity, between what has been
previously perceived and learned (i.e. schemata) and
new information. Cognitive dissonance theory tells us
that we try to reduce or avoid the discomfort.23 Clearly,
dissonance is a common occurrence as we interact
with our environment, especially with our peer group,
and the intensity of the dissonance depends on
how we perceive the competence of the person or
group expressing the contradictory opinion and our
emotional relationship with the person or group. It is
fundamentally the result of a reflective process, which
may involve either reflection in action24 or delayed
reflection. How we react when we face dissonance falls
into the following responses:
Learning support for the consultation • P Robinson et al. 431
� Blackwell Publishing Ltd MEDICAL EDUCATION 2003;37:429–433
1 we dismiss the message (story) as being unimportant
or wrong;
2 we dismiss the messenger (narrator) as unimportant;
3 we seek corroboration in the environment in support
of the new information and incorporate it into the
personal schema, or
4 we attempt to reduce the dissonance by negotiating
with the messenger, or, if this is not possible, we
accept the message and change the personal schema.
The easiest responses are types 1 and 2, especially if
we have developed schemata to support these behav-
iours. However, we learn nothing with such rigid
responses and can only learn with responses 3 and 4.
Cognitive dissonance is at the root of all learning and
strategies to induce dissonance, or �learning-by-disturb-
ing�25 are important. Practical support is needed to
optimise responses 3 and 4.
Reflection
Activity and reflection are two integral parts of the
learning process.10 In our context, �activity� is the
awareness of experience. Reflection may occur at
different times in relation to the activity. Schon24
describes how expert practitioners are able to reflect in
action. This process is often subliminal: it works on
automatic pilot. At higher levels of challenge the
practitioner may need to stop and think, perhaps
deferring a decision until this has been accomplished.
Schon also relates how an intuition that �something
does not fit� may not interrupt the practitioner’s flow,
but may be resolved on later reflection. Eraut26
describes in teachers the ability to store �film clips� of
classroom episodes. These may be replayed weeks or
months later and provide material for reflection.
We have considered how cognitive dissonance arises
when schemata pertaining to the same event conflict
with each other. Reflection (�the reconstruction of
experience�)9 is the process by which schemata are
compared, and so may come into conflict. This
reflection, and therefore dissonance, may arise in
action, during a pause in the action, shortly afterwards
or a long time later.
Implications for learning
We have described ways in which the GP’s experience
is socially determined and represented in cognitive
constructs (schemata), which contain narrative and are
expressed in narrative. If research-based knowledge is
to be used by a practitioner, it must become part of his
or her experience. The challenge of getting research
findings into practice becomes the task of incorporating
new information into the practitioner’s existing sche-
mata, and developing schemata to utilise learning
support appropriately. In other words the task is to
help the practitioner to learn. Further, the learning
required is situated in the consultation. How is this
learning to be supported?
Part of the solution is to provide information at the
point of care.27 If this information is in conflict with the
practitioner’s experience, it will lead to dissonance.
Cognitive dissonance is a necessary precursor of learn-
ing, but it can also lead to emotional tension and
avoidance of learning. In the nature of reflection, the
dissonance that occurs may be appreciated at the time
or later, during the consultation or after it. Learning
can take place outside a consultation, within a com-
munity of practice, as long as the story is simulated and
related to the consultation. In this situation the patient
is represented by an instance (case memory), and the
evidence by an illness story. The learning consists of
bringing the two stories together.
Gaining expertise implies the development of spe-
cialised schemata that frame perception and tend to
discard new information that does not match experi-
ence. In an unchanging world that had itself become
highly specialised, this would not be problematic.
However, in a complex, generalised environment where
the domain knowledge changes rapidly, there are issues
of updating schemata. Of course the brain is itself a
complex system and highly adaptive – what is required
are schemata that prepare us to constantly challenge
personal experience. These schemata need to be part of
professional culture, and developed in the community
of practice. They should also be general purpose
schemata that the individual can adapt to their own
circumstances. This type of schema is often best
communicated by metaphor. For example, the meta-
phor of education as a journey is a general purpose
schema that explains and supports personal change.
Conclusion
This paper is written from the point of view of Primary
Care in the UK. The issues discussed apply to the use
of on-line decision support in any medical context.
In the information age, what identifies the professional
is not information, as this is ubiquitous, but experience.
Traditionally, the doctor has been seen, or has aspired to
be seen, as a scientist who uses logical analysis and
hypothetico-deductive reasoning to solve clinical prob-
lems. In fact, hypothetico-deductive reasoning is not a
cognitive process but a schema, and we have tried to
move forward by replacing this view with a model of
Learning support for the consultation • P Robinson et al.432
� Blackwell Publishing Ltd MEDICAL EDUCATION 2003;37:429–433
experience that takes account of cognitive, social and
affective factors, and that identifies the learning methods
that enable evidence to get into practice. To make the
best use of what computerised decision support has to
offer, the individual practitioner is required to make use
of a variety of schemas, of which hypothetico-deductive
reasoning is only one. The link between these factors lies
in narrative, and we believe that learning support in the
consultation should reflect this.
As well as providing information at the point of care,
effective learning support depends on the practitioner
having schemata that help him or her to make positive
use of cognitive dissonance. Learning support needs to
provide practitioners with these schemata. It needs to
present new information in a way that takes account of
the practitioner’s experience and narrative structures. It
also needs to support learning at different stages in the
consultation and its aftermath.
Contributors
The ideas for this paper were developed by IP and PR.
RW provided details of the PRODIGY evaluation. All
three authors were involved in preparing the text.
Acknowledgements
We thank Nick Booth, Alison Evans and Phillip
Heywood, who all provided thoughtful feedback on
early drafts of this paper. We are also in debt to those
colleagues with whom we have exchanged ideas about
education and decision support.
Funding
None.
References
1 Purves IN. Concepts in Health Informatics. In: Simpson LC,
Robinson PJ, eds. E-Clinical Governance: A Guide for Primary
Care. Abingdon: Radcliffe Medical Press 2002; pp. 11.
2 Delaney BC, Fitzmaurice DA, Riaz A, Hobbs FDR. Can
computerised decision support systems deliver improved
quality in primary care? BMJ 1999;319:1281–3.
3 Purves IN. PRODIGY. Implementing clinical guidance using
computers. Br J Gen Pract 1998;48:1552–3.
4 Purves IN. The changing consultation. In: van Zwanberg T,
Harrison J, eds. GP Tomorrow – Living with Uncertainty.
London: Radcliffe Press 1999; pp. 31–49.
5 Robinson PJ, Heywood P. What do GPs need to know? The
use of knowledge in general practice consultations. Br J Gen
Pract 2000;50:56–9.
6 Wilson RG, Thomas JM. PRODIGY Workshop Summary
Report. PRODIGY Report 31. Sowerby Centre for Health
Informatics, School of Health Sciences, University of
Newcastle 1998; http://www.prodigy.nhs.uk/main.htm
7 Dowie J. The research practice gap and the role of decision
analysis in closing it. Health Care Anal 1996;4:1–14.
8 Schmidt HG, Norman GR, Boshuizen PA. A cognitive per-
spective on medical expertise: theory and implications. Acad
Med 1990;65:611–21.
9 Dewey J. How We Think – a Restatement of the Relation of
Reflective Thinking to the Educative Process. Boston: Heath 1933.
10 Kolb D. Experiential Learning. Englewood Cliffs, New Jersey:
Prentice Hall 1984.
11 Popper K. Conjectures and Refutations: the Growth of Scientific
Knowledge. London: Routledge & Kegan Paul 1963.
12 Lave J, Wenger E. Situated Learning. Legitimate Peripheral
Participation. Cambridge: Cambridge University Press 1991.
13 Wenger E. Communities of Practice. Learning, Meaning and
Identity. Cambridge: Cambridge University Press 1998.
14 Wilson R. An inductive study of the PRODIGY facilitated groups.
Factors in the use of PRODIGY and their effect on the use and
non-use. MSc Thesis. University of Newcastle 1997.
15 Johnson PE. What kind of system should an expert be? J Med
Philos 1983;7:77–97.
16 Brookfield SD. Understanding and Facilitating Adult Learning.
Milton Keynes: Open University Press 1986.
17 Hunter KM. Doctors’ stories. The Narrative Structure of Medical
Knowledge. Chichester: Princeton University Press 1991.
18 Anderson RC. Role of the reader’s schema in comprehension,
learning and memory. In: Anderson RC et al., eds. Learning to
Read in American Schools. Hillsdale, New Jersey: Erlbaum 1984.
19 Kay S, Purves IN. Medical records and other stories: a nar-
ratological framework. Methods Inf Med 1996;35:72–88.
20 Mandler JM. Stories, Scripts and Scenes: Aspects of Schema
Theory. Hillsdale, New Jersey: Erlbaum 1984.
21 Mattingly C. The narrative nature of clinical reasoning. Am J
Occup Ther 1991;45 (11):998–1005.
22 Penrose R. The Emperor’s New Mind: Concerning Computers,
Minds and the Laws of Physics. Oxford: Oxford University Press
1989.
23 Festinger L. A Theory of Cognitive Dissonance. Stanford:
Stanford University Press 1957.
24 Schon D. The Reflective Practitioner. How Professionals Think in
Action. London: Maurice Temple Smith 1983.
25 Aı̈meur E. Application and assessment of cognitive-disson-
ance theory in the learning process. J Univ Comp Sci 1998;
4 (3):216–47.
26 Eraut M. Developing Professional Knowledge and Competence.
London: Falmer Press 1994;28.
27 Sullivan F, Gardner M, van Rijsbergen K. An information
retrieval system to support clinical decision making at the
point of care. Br J Gen Pract 1999;49:1003–7.
Received 13 October 2000; editorial comments to authors 22 February
2001, 6 October 2002; accepted for publication 20 November 2002
Learning support for the consultation • P Robinson et al. 433
� Blackwell Publishing Ltd MEDICAL EDUCATION 2003;37:429–433