1
Resource 0 H~court Braceand CompanyLtd 1998 Clinical Investigations and Statistics in Laboratory Medicine R G Jones and R B Payne. ACB Venture Publications 1997, ISBN 0-902429-21-3, 192 pp, £21.00 Let me declare an interest. One of the authors is an old pal, the book is dedicated to another who died untimely, and I was privileged to comment on an early draft I am going to tell you that this is a terrific book which might have a much more interesting title like 'Statistics for fun.' So one simple example to demonstrate that this might be worth having on your shelves. There is an 'envoi' on p 188, with a quote from Alan Maynard: Until more is known about what works and why practitioners change practice, all NHS facilities and the Department of Health should exhibit large signs saying: In God we trust. All others bring data. Richard Jones and Brian Payne have managed to write a statistics book that is at once human and informative. It starts with an assumption that most of us know nothing, and that what little we knew we have forgotten. They give some enchanting one- page backgrounds to five famous statisticians: Gauss, Fisher, Deming, Bradford Hill and Cochrane, which makes them real people and their subject ours. Though written for laboratory medicine, actually this book is a gentle, hand-holding, informative and innovative entrance to health care research and practice. Yes there is a concentration on the sorts of things that go on in laboratories, and therefore a concentration on diagnostic testing rather than on outcomes. But making a diagnosis is where you start, after all. Laboratories consume large amounts of health care spending, and much of that, it is said (though without much evidence other than anecdote) is largely irrelevant. How will you know how relevant it is? Do you speak the language of sensitivity, specificity, normal ranges, log transformations or outliers? I suspect not. Yet diagnosis and the business of laboratories is likely to be where the next big push will come in the evolution of evidence- based health care. Systematic reviews and meta-analyses of treatments are beginning to be published in numbers, in journals, and available from the Cochrane Collaboration and the Centre for Reviews and Dissemination in York. We understand the ground rules, and we are beginning to apply the fruits of research in practice. Health economics is increasingly based on the rock of systematic review. But diagnosis and diagnostics are shifting sands, with much work to be done to firm up this ground. And diagnostic testing is an open door to technology creep, with new tests coming in, old tests being used in new areas, and more interventions being based on them. So getting to grips with the language and background now makes sense if we have to deal with the results of technology creep tomorrow. Laboratory medicine is a much bigger area than it ever used to be. New specialist laboratories spring up to do complex immunology, allergy testing, and increasingly aspects of genetics and screening. Yet tile basic principles of how to look at a test, whether a laboratory finding, or imaging, or a clinical observation, remain the same. This book will help understand the basic principles. It is actually more than that: a really useful primer on how to design a study, write a grant, write up the results or make a presentation. So it's practical as well as theoretical. It is useful to a wider audience than those stuck in a lab somewhere. It does have limitations. For instance, though it will show you (very nicely) how to calculate a likelihood ratio and use it (or an NNT for that matter), it is light on examples. That can be refreshing, because when you want to do either of these, you'll know just where to go for a simple memory boost. And that is where the main worth of this book lies. It can be a friend on your shelf, offering help, not a tome threatening to bash you with symbols and equations and ridicule your ignorance. Books like this are rare. Grab one. Andrew Moore Pain Research Oxford, UK 28 EVIDENCE-BASED HEALTtt POLICY AND MANAGEMENT MARCH 1998

Clinical investigations and statistics in laboratory medicine

Embed Size (px)

Citation preview

Resource 0 H~court Brace and Company Ltd 1998

Clinical Investigations and Statistics in Laboratory Medicine

R G Jones and R B Payne. ACB Venture Publications 1997, ISBN 0-902429-21-3, 192 pp, £21.00

Let me declare an interest. One of the authors is an old pal, the book is dedicated to another who died untimely, and I was privileged to comment on an early draft I am going to tell you that this is a terrific book which might have a much more interesting title like 'Statistics for fun.' So one simple example to demonstrate that this might be worth having on your shelves. There is an 'envoi ' on p 188, with a quote from Alan Maynard:

Until more is known about what works and why practitioners change practice, all NHS facilities and the Department of Health should exhibit large signs saying: In God we trust. All others bring data.

Richard Jones and Brian Payne have managed to write a statistics book that is at once human and informative. It starts with an assumption that most of us know nothing, and that what little we knew we have forgotten. They give some enchanting one- page backgrounds to five famous statisticians: Gauss, Fisher, Deming, Bradford Hill and Cochrane, which makes them real people and their subject ours.

Though written for laboratory medicine, actually this book is a gentle, hand-holding,

informative and innovative entrance to health care research and practice. Yes there is a concentration on the sorts of things that go on in laboratories, and therefore a concentration on diagnostic testing rather than on outcomes. But making a diagnosis is where you start, after all. Laboratories consume large amounts of health care spending, and much of that, it is said (though without much evidence other than anecdote) is largely irrelevant.

How will you know how relevant it is? Do you speak the language of sensitivity, specificity, normal ranges, log transformations or outliers? I suspect not. Yet diagnosis and the business of laboratories is likely to be where the next big push will come in the evolution of evidence- based health care. Systematic reviews and meta-analyses of treatments are beginning to be published in numbers, in journals, and available from the Cochrane Collaboration and the Centre for Reviews and Dissemination in York. We understand the ground rules, and we are beginning to apply the fruits of research in practice. Health economics is increasingly based on the rock of systematic review.

But diagnosis and diagnostics are shifting sands, with much work to be done to firm up this ground. And diagnostic testing is an open door to technology creep, with new tests coming in, old tests being used in new areas, and more interventions being based on them. So getting to grips with the language and background now makes sense if we have to deal with the results of technology creep tomorrow.

Laboratory medicine is a much bigger area than it ever used to be. New specialist laboratories spring up to do complex immunology, allergy testing, and increasingly aspects of genetics and screening. Yet tile basic principles of how to look at a test, whether a laboratory finding, or imaging, or a clinical observation, remain the same.

This book will help understand the basic principles. It is actually more than that: a really useful primer on how to design a study, write a grant, write up the results or make a presentation. So it 's practical as well as theoretical. It is useful to a wider audience than those stuck in a lab somewhere.

It does have limitations. For instance, though it will show you (very nicely) how to calculate a likelihood ratio and use it (or an NNT for that matter), it is light on examples. That can be refreshing, because when you want to do either of these, you'l l know just where to go for a simple memory boost. And that is where the main worth of this book lies. It can be a friend on your shelf, offering help, not a tome threatening to bash you with symbols and equations and ridicule your ignorance.

Books like this are rare. Grab one.

Andrew Moore Pain Research

Oxford, UK

28 EVIDENCE-BASED HEALTtt POLICY AND MANAGEMENT MARCH 1998