See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/24917811

Practical Statistics for Medical Research

Article  in  Journal of Epidemiology & Community Health · October 1992

DOI: 10.1136/jech.46.5.549-a · Source: PubMed Central

CITATIONS READS

23 4,217

1 author:

Robert G Newcombe
Cardiff University
266 PUBLICATIONS   11,717 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

UK-ANZ POSNOC trial. www.posnoc.co.uk View project

Adolescent sexual health View project

All content following this page was uploaded by Robert G Newcombe on 28 February 2014.

The user has requested enhancement of the downloaded file.

 549

unevenly, the responses to a series of ques- The book is intended for the "users" of "Preparing to analyse data", "Clinical
tions about policies and range of services: research rather than the "doers" and is organ- trials", and "The medical literature". Each
organisation and funding, historical back- ised in terms of clinical actions-diagnosis, chapter describing analytical methods
ground and current details of home care, management, and keeping up to date. The includes a section on the presentation of
service evaluation, and policy debate. style is that of a self instruction manual, results.
There are chapters from Belgium, Den- including six aide memoire cards for the Douglas Altman is one of the statisticians
mark, France, Germany, Italy, The Nether- white coat, presumably to be whipped out on responsible for the very welcome shift in
lands, the United Kingdom, and curiously, ward rounds or consulted surreptitiously at emphasis in leading medical journals away
Israel. Coverage varies greatly. The Belgian less opportune moments. The second edition from significance testing and towards esti-
chapter limits itself largely to arrangements in is 70 pages longer than the first and mation and confidence intervals. The
Flanders, themselves sufficiently complex to incorporates new material on concepts such emphasis in this book corresponds closely to
deter all but the most determined in their as numbers needed to treat, n of 1 trials, this, leaving room for criticism only by those
curiosity about what happens in the other half interpretation of confidence intervals, and who reject even the confidence interval
of the country. For Italy read Genoa, and not quality of life measures, as well as a major approach in favour of Bayesianism.
even very much about that. Danish services, revision of the 120 page section on keeping up The author is concerned to establish use of
organised by small local government units, to date. This section now includes instruc- terminology that will not mislead: thus
are evidently uniform, broadly adequate, and tions on how to do literature searches, both "reference interval" is preferred to "normal
successful in their main aim of avoiding traditional and electronic. How much this has
unnecessary institutionalisation. to do with epidemiology is debatable but I did range", and regression involves outcome and
It would be unkind to treat this serious wonder whether British clinicians will be predictor variables, not dependent and
attempt to grapple with the diversity of persuaded to use Medline to solve their independent ones.
Useful examples of displaying data that
origins, achievements, and challenges in clinical problems. In contrast the excellent make the most of the high resolution graphics
home care as a kind of Eurovision Home Help sections on diagnosis and management are a now available include a plot with several dot
Contest, but inevitably comparisons arise. superb exposition of the application of epi-
There is a broad distinction between two demiological expertise to clinical issues such diagrams and summary statistics for each
column of points (page 40) and a scattergram
traditions; one of charity, the other of entitle- as diagnostic testing, screening, and matrix depicting graphically, but compactly,
ment, with complexity and unevenness attribution of adverse events. all the elements of a correlation matrix (page
flourishing in the wake of the former. The There are now at least four North
casual Eurovision viewer, especially if old, American texts with the title "Clinical epi- 342).
might be forgiven for favouring the latter. demiology", some testimony perhaps to the Each chapter includes exercises, which are
All services face problems relating to vigour of clinical epidemiology there. Who quite searching and designed to be instruc-
referral and assessment, to organisation, should read this particular book? Certainly tive, involving both performing analyses and
funding, and evaluation. Home care-even clinicians, to whom it can be strongly interpretation. Helpful explanatory solutions
are given. A small minority of the exercises
without a recession- is inherently difficult: a recommended. I have also found it a valuable
social endeavour comparable to that greatest source of material for undergraduate and have somewhat surprising contexts
of military challenges, the fighting retreat. postgraduate teaching. Public health physi- (astronauts, male singers)-in the latter case
The old get older and frailer too. There are cians may prefer the more concise approach of the relevance of the data to the issue being
crises and periods of usually temporary relief; Fletcher, Fletcher and Wagner's Clinical epi- tested seems far fetched. Another exercise, on
there is a need for constant vigilance and demiology. aviation accidents, is unclear, but apart from
occasional energetic improvisation. But the these, the general standard of the exercises is
surrender to institutionalisation, usually very high.
expensive and almost always unwelcome, is RICHARD LOGAN The bibliography section indicates the
not one that comes easily. Department of Public Health Medicine and page in the text where each reference is
This book, with all its faults, at least Epidemiology cited-a practice that should become
succeeds in portraying on a European scale Queen's Medical Centre standard.
the complexity of the problems involved. It Nottingham According to the preface, the book is
underlines common aspirations and some recommended for medical researchers who
common means. We are not alone in our have had some statistics teaching a long time
uncertainties about home care. And care ago, and is also useful for students, clinicians,
management, however difficult in practice, is and those doing postgraduate courses. The
the best means we have of reducing them. Practical Statistics for Medical present price puts the book beyond the reach
Research. By Douglas G Altman. (Pp 611; of medical undergraduates, though medical
COLIN CURRIE £32.) London: Chapman and Hall, 1991. school libraries should be encouraged to have
Department of Geriatric Medicine ISBN 0-412-27630-5. several copies available. Provision of a more
City Hospital reasonably priced paperback edition would
Edinburgh This excellent book is designed to instil a put this helpful book within the reach of many
thoughtful approach into the reading of more readers. The book is of great value to
research literature and the planning and specialists in public health medicine and
interpretation of research studies. The author MFPHM trainees, and includes aspects of
is concerned to develop the reader's insight study design relevant to epidemiology, and
into the fundamental concepts of statistics as confidence intervals for measures of relative
Clinical Epidemiology: A Basic Science applied to medicine, and how they fit into the
for Clinical Medicine By D L Sackett, R B reality of performing and assessing research. risk. The scope is exclusively clinical and
Haynes, G H Guyatt, and P Tugwell. (Pp In the first chapter, inadequacies of presenta- epidemiological research, however, and
400; C23.95.) Boston: Little Brown, 1991. tion of results by the media and manu- trainees tend to express a demand also for
ISBN 0-316-76599-6. books such as that of Woodward and Francis1
facturers are presented in such a way that the in which the orientation is towards health
reader is led to consider how data should be services management.
A concise definition of clinical epidemiology obtained. The objective is not to present a
remains elusive; indeed its existence as a catalogue of methods, still less one of
distinct branch of epidemiology has been a formulae-but the latter are regarded as help- R G NEWCOMBE
source of much contention. The authors of ful in showing how analyses work, and it is Department of
this book, who are all practising clinicians, recommended that except in cases of extreme Medical Computing and Statistics
regard clinical epidemiology as a basic science hypersensitivity the reader should seek to University of Wales
for clinical practice and define it as being understand how they apply. College of Medicine Cardiff.
what clinical epidemiologists do. While their The careful approach is exemplified by
pragmatic approach has attractions it leaves some of the chapter headings: these include 1 Woodward M, Francis LMA. Statistics for
some important issues of content unresolved. "Designing research"-of course but also, health service management and research.
London: Edward Arnold, 1988.

View publication stats