01.06.2013 Views

Statistical Methods in Medical Research 4ed

Statistical Methods in Medical Research 4ed

Statistical Methods in Medical Research 4ed

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

2 The scope of statistics<br />

other patient. The cl<strong>in</strong>ical judgement exercised by a physician <strong>in</strong> the choice of<br />

treatment for an <strong>in</strong>dividual patient is based to an extent on theoretical considerations<br />

derived from an understand<strong>in</strong>g of the nature of the illness. But it is<br />

based also on an appreciation of statistical <strong>in</strong>formation about diagnosis, treatment<br />

and prognosis acquired either through personal experience or through<br />

medical education. The important argument is whether such <strong>in</strong>formation should<br />

be stored <strong>in</strong> a rather <strong>in</strong>formal way <strong>in</strong> the physician's m<strong>in</strong>d, or whether it should<br />

be collected and reported <strong>in</strong> a systematic way. Very few doctors acquire, by<br />

personal experience, factual <strong>in</strong>formation over the whole range of medic<strong>in</strong>e, and it<br />

is partly by the collection, analysis and report<strong>in</strong>g of statistical <strong>in</strong>formation that a<br />

common body of knowledge is built and solidified.<br />

The phrase evidence-based medic<strong>in</strong>e is often applied to describe the compilation<br />

of reliable and comprehensive <strong>in</strong>formation about medical care (Sackett et<br />

al., 1996). Its scope extends throughout the specialties of medic<strong>in</strong>e, <strong>in</strong>clud<strong>in</strong>g, for<br />

<strong>in</strong>stance, research <strong>in</strong>to diagnostic tests, prognostic factors, therapeutic and prophylactic<br />

procedures, and covers public health and medical economics as well as<br />

cl<strong>in</strong>ical and epidemiological topics. A major role <strong>in</strong> the collection, critical evaluation<br />

and dissem<strong>in</strong>ation of such <strong>in</strong>formation is played by the Cochrane Collaboration,<br />

an <strong>in</strong>ternational network of research centres (http://www.cochrane.org/).<br />

In all this work, the statistical approach is essential. The variability of disease<br />

is an argument for statistical <strong>in</strong>formation, not aga<strong>in</strong>st it. If the bedside physician<br />

f<strong>in</strong>ds that on one occasion a patient with migra<strong>in</strong>e feels better after dr<strong>in</strong>k<strong>in</strong>g<br />

plum juice, it does not follow, from this s<strong>in</strong>gle observation, that plum juice is a<br />

useful therapy for migra<strong>in</strong>e. The doctor needs statistical <strong>in</strong>formation show<strong>in</strong>g,<br />

for example, whether <strong>in</strong> a group of patients improvement is reported more<br />

frequently after the adm<strong>in</strong>istration of plum juice than after the use of some<br />

alternative treatment.<br />

The difficulty of argu<strong>in</strong>g from a s<strong>in</strong>gle <strong>in</strong>stance is equally apparent <strong>in</strong> studies<br />

of the aetiology of disease. The fact that a particular person was alive and well at<br />

the age of 95 and that he smoked 50 cigarettes a day and drank heavily would not<br />

conv<strong>in</strong>ce one that such habits are conducive to good health and longevity.<br />

Individuals vary greatly <strong>in</strong> their susceptibility to disease. Many abstemious<br />

non-smokers die young. To study these questions one should look at the morbidity<br />

and mortality experience of groups of people with different habits: that is,<br />

one should do a statistical study.<br />

The second chapter of this book is concerned ma<strong>in</strong>ly with some of the basic<br />

tools for collect<strong>in</strong>g and present<strong>in</strong>g numerical data, a part of the subject usually<br />

called descriptive statistics. The statistician needs to go beyond this descriptive<br />

task, <strong>in</strong> two important respects. First, it may be possible to improve the quality<br />

of the <strong>in</strong>formation by careful plann<strong>in</strong>g of the data collection. For example,<br />

<strong>in</strong>formation on the efficacy of specific treatments is most reliably obta<strong>in</strong>ed<br />

from the experimental approach provided by a cl<strong>in</strong>ical trial (Chapter 18),

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!