Article Text


Research methods in occupational epidemiology, 2nd edition
  1. D Coggon

    Statistics from

    Havey Checkoway, Neil E Pearce, and David Kriebel. Oxford: Oxford University Press, 2004, (£29.95, hardback), ISBN 0-19-509242-2

    When this book was first published, it rapidly became a standard reference for occupational epidemiologists, and for some years it has been the leading textbook in its field. A new edition, extensively revised and updated, is therefore most welcome. Changes include the addition of sections on case-cohort and case-crossover designs, and on the statistical analysis of repeated measures data, as well as the incorporation of many practical examples from more recently published research to illustrate theoretical points. There is also a new chapter on epidemiological surveillance of occupational hazards.

    A good test of an epidemiological textbook is the way in which it covers the difficult topics of confounding and the case-control method, and on both these counts the book is a winner. The principles of each are clearly and logically developed. Indeed, the clarity of the text throughout is to be commended. Another example is the way in which the authors point out that once a study has been completed, power calculations become superfluous because the potential impact of sample size on statistical uncertainty can be much more meaningfully summarised by the confidence intervals around effect estimates. Unfortunately this message has still not got through to some researchers.

    A further strength is the presentation of general estimating equations as an overarching framework that embraces commonly used analytical techniques such as linear, logistic, and Poisson regression. While many readers will find the detailed mathematics of these methods beyond them, and those who want to understand them in depth will need to refer to other sources, placing them in their logical context in this way is helpful.

    At 372 pages, the book can be read from cover to cover, and does not serve only as a reference. This brevity and readability does, however, mean some restriction of scope, and readers will derive more from it if they are first familiar with the basic principles of epidemiology more generally. For example, there is little discussion of the practical challenges of case definition and approaches to their validation, and the development of concepts such as incidence, prevalence, and the various measures of association between exposure and disease may be a little cursory for the uninitiated reader. And unsurprisingly, the focus leans towards the authors’ particular areas of interest, with an emphasis on studies of mortality and cancer incidence in relation to chemical and physical exposures, and relatively less on reproductive outcomes, musculoskeletal disorders, or the assessment of ergonomic and psychosocial risk factors.

    As always, there are a few minor criticisms that one can make. In more than 20 years experience as an occupational epidemiologist in the UK, I have never come across the “Central Record Office of the Ministry of Pensions and National Insurance”, and the authors are rather dismissive of period prevalence, although for some health outcomes such as back pain and finger blanching it may be the most appropriate measure to use.

    Overall, however, this is a book that can be thoroughly recommended. Every department of occupational epidemiology should have a copy, and those teaching the subject should consider using it as part of their teaching material. Occupational health practitioners with a more peripheral interest in epidemiology may find it useful to dip into, and it is structured in such a way that the more advanced sections can be easily identified and skipped over.

    View Abstract

    Request permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.