Ophthalmic Research and Epidemiology: Evaluation and Application. Stanley W.Hatch, ed. Boston: Butterworth‐Heinemann, 2000. Pages: 298. Price: $47.50. ISBN 0‐7506‐9914‐0.

Ophthalmic Research and Epidemiology: Evaluation and Application. Stanley W. Hatch, ed. Boston: Butterworth-Heinemann, 2000. Pages: 298. Price: $47.50. ISBN 0-7506-9914-0. Lisa A. Jones, The Ohio State University College of Optometry, Columbus, Ohio According to the Preface of Ophthalmic Research and Epidemiology, the goal is to provide the basic tools for clinicians to design and interpret clinical research. This would assist them in finding and assimilating quality research into their practice. The basics of study design, analysis, and results interpretation are outlined to allow clinicians to do this. Overall, the book is successful in highlighting issues that are crucial for understanding and designing quality studies. The objective is ambitious, covering issues in almost every aspect. It could be an addition to the library of a clinician seeking to expand his knowledge in the research arena that is at a loss for where to begin. Each chapter has several different aids to enhance key points. At the beginning of the chapter is a brief behavioral objective. Clinical pearls and key concepts within the chapter provide reinforcement for important clinical and research information. The key concepts are very helpful because they provide a succinct explanation of ideas that the author finds particularly relevant. Each chapter ends with a self-assessment quiz with answers supplied and references for further reading. Both are useful additions that supplement the outline the book provides. The beginning of the book describes the purpose of clinical research and provides an overview of study design types. Chapter 3 is titled Cause and Effect; it addresses issues about causal relationships and what strengthens the evidence for them. There is a brief discussion of bias (selection, information, and confounding) and how it affects the determination of a causal relationship. It brings necessary attention to concepts that are often not considered, such as biological plausibility and strength of association. The next two chapters deal with research articles. The first gives suggestions for reading and critically analyzing articles. It introduces another interesting learning tool, using journal abstracts to exemplify a point being addressed. The next chapter outlines the sections of an article (title, abstract, etc.). I think that a discussion of authorship, outside of acknowledgments, would be an extremely useful addition given current ethical concerns about authorship. The next four chapters (chapters 6 to 9) are the core of the design and analysis section. The Principles of Study Design chapter discusses study validity and revisits the bias discussion. Detail is given about study design types (case-control, cohort, etc.), when to use them, and the advantages and disadvantages. In the randomized clinical trial section, entry criteria, randomization, and types of sampling schemes are presented. At the end of the chapter, there is a rather detailed section on meta-analysis and decision analysis. An introduction to these is good because use of these in the medical literature is growing, but perhaps too much information is provided. Descriptive Statistics (chapter 7) and Measures of Association (chapter 8) present ways of communicating basic information about data. Central tendency and variability measures (mean, variance, etc.) are described, with formulas to allow for calculation. Examples to demonstrate how the formula works are provided in each case to aid in understanding. Measures of disease frequency, such as incidence rates, are presented in detail with formulas and examples. Definitions for other less common epidemiologic terms are given. The chapter on statistics contains a handy table (Table 9-1) for quick reference about which statistical test to use. Parametric methods for data are presented along with how to calculate them. Nonparametric methods are discussed as well as the important concept of effect modification (also known as interaction). Information of the more advanced analytic techniques, such as linear and logistic regression, are discussed as well. The chapter is fairly comprehensive in presenting the most common statistical methods. There is a difficult line to walk when determining whether to present a working knowledge of the methods or whether the reader should be able to use the book to actually do analysis. Some of the sections could have been pared down (most notably the analysis of variance section) in terms of the amount of the detailed information they contain. The remaining chapters attack different issues. The legal and ethical issues of conducting research, including the extremely important topic of informed consent, are tackled. This chapter (Chapter 10) is noteworthy in its lack of discussion about institutional review boards, a necessary hurdle for most types of research in academic and many medical settings. The factors involved in developing and implementing a screening program and determining its effectiveness are discussed in Chapter 11. Concepts such as sensitivity and specificity are defined and demonstrated. Survey research methods are the topic of another chapter. There are sections on the use of surveys, methods of administration, testing a survey, and ways to improve response rates. Many examples of studies using surveys and published surveys are given. The author set quite a task for himself. The breadth of information to be covered under the topic is extensive. Given that, it is easy to say that there are things that should have been included and things that should have been excluded. Aside from that, there are things with which I disagree. For example, on page 125 there is a discussion about two confidence intervals overlapping. This is not a way to distinguish whether two independent samples are different. I also disagree with the statement made about case-control studies providing an unbiased estimate of relative risk or incidence rate ratio (page 83). Although it provides an estimate, it is probably rare for a case-control study to overcome all sources of bias to produce an unbiased estimate. It is not appropriate to assume that this is true. There are also instances of incompleteness. For example, the definition and example of differential misclassification is limited to measurement devices. In reality, differential misclassification can occur in situations other than when devices are used. The definition of confounding on page 159 is does not provided sufficient information. Overall, the information has been presented in a basic form that should be accessible to clinicians with little experience in research matters. Given what the author set out to do, I believe that he was true to the goals outlined in his preface.