Journal of Evaluation in Clinical Practice, 8, 2, 127–130

Data, information and knowledge: the health informatics model and

its role in evidence-based medicine
Andrew Georgiou BA DipArts MSc
Coronary Heart Disease Programme Co-ordinator, Clinical Effectiveness and Evaluation Unit, The Royal College of
Physicians, London, UK

Correspondence Abstract
Mr Andrew Georgiou The health informatics model consists of three essential parts: data, infor-
CHD Programme Co-ordinator
Clinical Effectiveness and Evaluation
mation and knowledge. These elements are arranged in a hierarchy, with
Unit data at the base of the model providing the basis for establishing informa-
The Royal College of Physicians tion and leading in turn to the potential generation of knowledge. The
11 St Andrew’s Place informatics model converges closely with the principles, aims and tasks of
London NW1 4LE
UK
evidence-based medicine (EBM), particularly as they relate to searching,
E-mail: [email protected] appraising, reviewing and utilizing information and research. The develop-
ment of health informatics today has its origins in the growth of statistics
Keywords: evidence-based medicine, in the 18th and 19th centuries. As a new and growing discipline, statistics
health informatics, health care
burgeoned amidst the challenge of measuring, monitoring and ultimately
management, information technology,
statistics governing societies in the throes of massive change and expansion. The
governance role embraced by statistics in the past resembles many aspects
Accepted for publication: of the role ascribed to audit, quality assurance and EBM today. There are
8 January 2002
some deep-seated paradoxes within the field of health informatics. The
informatics model posits an oversimplified and linear progression of data
to information and knowledge. Health informatics may involve the spread-
ing and dissemination of information but this should be seen as only a part,
not the equivalent, of the complex process of generating knowledge.

Electronic Library of Health. Underlying the concept

Introduction
of EBM is a commitment to the establishment of
Evidence-based medicine (EBM) became a feature robust clinical information systems to help realize
of medical and health care planning in the 1990s. Sup- EBM goals (Muir Gray 1997).
porters of EBM explain that it is underpinned by a
shift in the culture of health provision away from
The rise of health informatics
decisions based on opinion, past practices and prec-
edent, towards a system that better utilizes science, While EBM’s growing acceptance and utilization has
research and evidence to guide decision making been aided by rapid IT developments, the use of IT
(Appleby et al. 1995). within the health service remains sporadic, uneven
Advances in information technology (IT) have and heavily under-resourced (Goodman 1998). Up
spurred the development of EBM, as well as provid- until recently computers and IT remained within the
ing the very basis for its realization (Georgiou 2001). background of medicine, often found within financial
These include the appearance of numerous easily offices, medical record offices, clinical laboratories
accessible databases, and powerful clinical search and research facilities, or else found in ‘niche systems’
engines and resource tools like the National like electrocardiographic analyses and patient moni-

© 2002 Blackwell Science 127

A. Georgiou

toring, or diagnostic instruments such as computer

tomographic scanners or magnetic resonance imag-
ing units. The common characteristic of all these ap- Knowledge
plications is that they are task-oriented and centred
mostly in particular departments or services. The
Information
major weakness of health care computing has con-
tinued to be the dichotomy between the abundant
level of computing expertise available on the one
Data
hand, and the lack of commensurable utilization of
computing on the other.
The importance of informatics in health care is Figure 1 The informatics model.
more than just a consequence of rapid IT develop-
ments. It is uniquely positioned at the intersection
of information technology and the many different
disciplines involved in, or associated with, medicine. sure or breathing may mean little, but when they are
Information processing and communication are cen- placed into the contextual framework of managing
trally involved in virtually all health care activities, a patient the data take on a meaningful existence.
including: obtaining and recording information about While data can include things that are known like
patients; communication among health care profes- facts or figures, information is usually assumed to be
sionals; accessing medical literature; selecting diag- the product of processed data. This often means that
nostic procedures; interpreting laboratory results; they have been manipulated in some way to ensure
and collecting clinical research data. their value. There is a parallel between the infor-
Health informatics can be defined as the discipline matics model and the principles and tasks of EBM,
that integrates biomedical sciences, computer sci- particularly as they relate to searching, appraising,
ences and health care policy, management and orga- reviewing and utilizing information and research, i.e.
nization (Peel 1994). transforming data and information into evidence-
Like the science of medicine, informatics is essen- based knowledge.
tially heterogeneous, and cannot escape the method- The informatics model is a simplistic way to con-
ological and epistemological issues involved in the ceptualize a complex process. It is also founded on
practice of medicine. some questionable assumptions, particularly if the
process of data/information/knowledge is viewed lin-
early, whereby the mere capture of data on one side
The health informatics model
of the spectrum can lead seamlessly to information
The informatics process traces its origins to the func- and then knowledge. In reality there are a number
tions of taxonomy and classification as they devel- of interrelated activities involved in the generation
oped in the 19th century. Early statisticians used and of information. These include the most demanding
developed classifications systems as ‘repositories of attention to data quality – validity, reliability, mean-
knowledge’ established from data and information ingfulness and accessibility – alongside careful regard
(Desrosieres 1998). They in turn developed an infor- of the statistical and epidemiological meaning of
matics model consisting of three essential parts the results. The generation of knowledge proceeds
arranged hierarchically, with data at the bottom, an through a complex process of induction, deduction
intermediary layer of information, and topped by the and assessment, itself subject to scientific debate and
knowledge layer (see Fig. 1). further trials and experimentation. The mere collec-
Within this model, data take on the character of tion of data and their presentation using impressive
facts or observations, which in and of themselves mathematical models are not a guarantee of useful
have little or no meaning. It is information that pro- information in the short term or of knowledge in the
vides the context for managing data. For instance, a long term. Moreover, the whole process cannot be
temperature reading or a figure denoting blood pres- divorced from the social, economic and even politi-

128 © 2002 Blackwell Science, Journal of Evaluation in Clinical Practice, 8, 2, 127–130

 Data, information and knowledge

cal influences that impact upon any decision-making that are interesting. These studies are often the ones
exercise. that are well funded and are submitted and published
The similarities between early 19th century de- more rapidly (Sutton et al. 2000).
velopments and today’s information agenda do not Governments and health insurers are attracted
end there. Desrosieres’s (1998) The Politics of Large to the notion of allocating ‘cost-effective’ health
Numbers: A History of Statistical Reasoning, a resources on the basis of ‘evidence’. There are many
description of the information model adopted in difficulties involved in resource allocation, however,
the 19th century, revealed a similarly constructed and often the large quantities of trial data required
circular model where ‘data’ (literally ‘givens’) appear to meet the standards of evidence just do not exist.
as a result of organised action; ‘information’ is the The effect of this may mean that EBM is introducing
result of the formatting and structuring of these a systematic bias towards treatments in areas where
data through nomenclatures and ‘knowledge’ and there are existing funds available to show effective-
‘learning’ result from the reasoned accumulation of ness (e.g. new pharmaceutical agents) at the expense
information. of other areas where rigorous evidence does not
currently exist or is not easily attainable (Kerridge
et al. 1998). Frankel et al. (2000) have complained that
various schemes proclaiming effectiveness, efficiency,
EBM: using ‘rationalism’ as a means
equity and clinical standards, masked a profoundly
of rationing?
unscientific rationing debate dominated by politics,
EBM has made dramatic inroads into many levels assertion and ethics.
of clinical practice and management. The discussion The strategies of the pharmaceutical industry are
about its value touches on fundamental questions of likely to result in more research being carried out in
epistemology: where does medical knowledge comes those areas where a pharmaceutical company has a
from and where ought it to come from? Is knowledge vested interest. This could lead to the situation where
based on an open scientific discourse or is it oligar- some drug treatments are recommended and pur-
chic and closed, based solely on expertise (Marshall chased, not because they are better than alternative
1997)? There are some who criticize EBM as treatments, but because the current existing evidence
statistically driven rather than scientifically driven, of effectiveness is better. Moreover, it is possible that
and complain that the health service has been too much reliance by purchasers on EBM may lead
‘forcibly unified under a single “quality assurance” to treatments being unavailable because they do not
system – easily regulated by politicians, bureaucrats represent the ‘best buy’ (Hope 1995).
and their statistical technicians’ (Charlton & Miles
1998, p. 373). Such criticisms contend that EBM
Conclusion: the health information paradox
threatens to replace healthy scientific debates,
where no one can claim authority over truth, with There can be no doubt that the ‘information age’ has
‘authorities’ of knowledge and readers of literature profoundly affected the ways and means by which
who announce a verdict and ensure that it is then society communicates, plans and organizes itself. IT
executed (Shahar 1997). will play a pervasive role in the health care environ-
Grimley Evans has raised questions about the ment of the future; the real question is whether
ability of large data bases to provide best guidance future systems are designed and implemented effec-
for clinical practice by pointing to the potential for tively to optimize technology’s role as a stimulus and
the misapplication of evidence-based medicine using support for the health care system and individual
irrelevant or outdated ‘evidence’ from randomized practitioners.
controlled trials, systematic reviews and expert Health informatics lies at the heart of medicine
guidelines (Grimley Evans 1995). There is therefore and health care itself, integrally bound up with the
an implicit bias against older people. Selection bias is process of clinical decision-making. However, its role
also a likely factor affecting meta analysis of trials. is not separate from the many areas of medical
Medical journals use studies with significant results science that it serves. At the bottom there are some

© 2002 Blackwell Science, Journal of Evaluation in Clinical Practice, 8, 2, 127–130 129

A. Georgiou

deep-seated paradoxes within the field of health Desrosieres A. (1998) The Politics of Large Numbers:
informatics. It is responsible for spreading and dis- a History of Statistical Reasoning. Harvard University
seminating information but this cannot be a ‘quick Press, Boston, MA.
and easy’ ‘cost-effective’ substitute for the process of Frankel S., Ebrahim S. & Davey Smith G. (2000) The limits
to demand for health care. British Medical Journal 321,
generating knowledge. The informatics model con-
40–45.
ceptualizes a linear and hierarchical progression of
Georgiou A. (2001) Health informatics and evidence-based
data to information and knowledge. This parallels
medicine – more than a marriage of convenience? Health
key EBM concepts, particularly as they relate to Informatics Journal 7 (3–4), 127–130.
ensuring the accessibility of evidence as a basis for Goodman K.W. (1998) Bioethics and health informatics:
clinical decision-making. However, the process of an introduction. In Ethics, Computing and Medicine (ed.
generating knowledge does not simply proceed K.W. Goodman), pp. 1–31. Cambridge University Press,
seamlessly through a measured statistical review of Cambridge.
currently existing trials or appraisal of existing Grimley Evans J. (1995) Evidence-based and evidence-
journal articles. It is a much more involved process, biased medicine. Age and Ageing 24, 461–463.
a product of intense scientific debate, successes and Hope T. (1995) Evidence based medicine and ethics.
failures. It rarely proceeds in increments but more Journal of Medical Ethics 21, 259–260.
Kerridge I., Lowe M. & Henry D. (1998) Ethics and
often in fits and starts and occasionally in leaps and
evidence based medicine. British Medical Journal 316,
bounds.
1151–1153.
Marshall T. (1997) Scientific knowledge in medicine: a new
Acknowledgements clinical epistemology? Journal of Evaluation in Clinical
Practice 3, 133–138.
The Royal College of Physicians of London provided Muir Gray J.A. (1997) Evidence Based Health Care.
an initial grant to the author for an MSc dissertation Churchill Livingstone, Edinburgh.
on which many of the ideas in this paper are based. Peel V.J. (1994) Management-focused health informatics
research and education at the University of Manchester.
References Methods of Information in Medicine 33, 273–277.
Shahar E. (1997) A Popperian perspective of the term
Appleby J., Walshe K. & Ham C. (1995) Acting on the ‘evidence-based medicine’. Journal of Evaluation in
Evidence a Review of Clinical Effectiveness: Sources of Clinical Practice 3, 109–116.
Information, Dissemination and Implementation. HSMC, Sutton A.J., Duval S.J., Tweedie R.L., Abrams K.R. & Jones
Birmingham. D.R. (2000) Empirical assessment of effect of publication
Charlton B. & Miles A. (1998) The rise and fall of EBM. bias on meta analyses. British Medical Journal 320,
Quarterly Journal of Medicine 91, 371–374. 1574–1577.

130 © 2002 Blackwell Science, Journal of Evaluation in Clinical Practice, 8, 2, 127–130