A Clinician’s Guide to Statistics and Epidemiology in Mental Health: Measuring Truth and UncertaintyRichard Balon, MD
Wayne State University, Detroit, MI, USA
By S. Nassir Ghaemi. New York, NY: Cambridge University Press; 2009; ISBN 978-0-521-70958-3; pp 151; $50 (paperback).
All statistics, as the author of this slender volume Dr. S. Nassir Ghaemi points out, is “an act of interpretation and the result of statistics is more interpretation” (p 3). Most clinicians (including myself) wrestle with understanding and, at times, doing statistics all through their career. Part of the difficulty is our understanding of this seemingly “different language of interpretation” (mine). Part of the difficulty is that we do not perform statistics on a daily basis (which reminds me of many other tasks, eg, ECG—unless one reads and interprets it daily, one does not do it well). Nevertheless, statistics should be and is a part of clinical practice. If one wants to understand the latest developments in medicine and psychiatry, he/she has to be able to understand the latest articles in medical journals and their statistical analyses. Also, as Dr. Ghaemi suggests, “It is thus not an option to avoid statistics, if one cares about science” (p 2). He also reminds us that statistics is not new: “Statistics were developed in the eighteenth century because scientists and mathematicians began to recognize the inherent role of uncertainty in all scientific work” (p 1). Yet, in spite of this, most of us keep having difficulty understanding statistical methods and interpretations. Dr. Ghaemi decided to write this book for those who have difficulty understanding statistics (not performing it!). “This is a book by a clinical researcher in psychiatry for clinicians and researchers in the mental health professions” (p xii).
The book, besides the Preface and Acknowledgements, is divided into 6 sections: 1. Basic concepts; 2. Bias; 3. Chance; 4. Causation; 5. The limits of statistics; and 6. The politics of statistics. The first section consists of 3 chapters—”Why data never speak for themselves,” “Why you cannot believe your eyes: The 3 C’s,” and “Levels of evidence.” In the first chapter the author points out that numbers do not stand alone and that facts always need to be interpreted; “…it is the job of statistics: not to tell us the truth, but to help us get closer to the truth by understanding how to interpret the facts” (p 4). The second chapter alludes to the fact that evaluation of any study needs to pass 3 hurdles—confounding bias, chance, and causation—before one considers accepting its results.
The second section includes another 3 brief chapters—”Types of bias,” “Randomization,” and “Regression.” The chapter on types of bias (confounding and measurement biases) emphasizes that “confounding bias is handled either by preventing it, through randomization in study design, or by removing it though regression models in data analysis” (p 13). This chapter includes several examples of confounding bias in published studies discussed in detail. In the next chapter on randomization Dr. Ghaemi states that randomization is the most revolutionary and profound discovery of modern medicine. It allows us, usually, to differentiate the true from the false, a real breakthrough from a false claim (p 21). The chapter on regression explains that another good way to reduce confounding bias in observational studies is stratification and regression, and discusses both these concepts. The titles of the 3 chapters of the third section—“Hypothesis-testing: the dreaded P-value and statistical significance,” “The use of hypothesis-testing statistics in clinical trials,” and “The better alternative: Effect estimation” pretty much explain what these chapters are about. The author again emphasizes the importance of randomization and tells us that it is the most important aspect of clinical trials, more important than placebo (the rationale for using placebo is to control for the natural history of the illness [p 58]). The fourth section (2 chapters) focuses on “What does causation mean?” and “A philosophy of statistics.”
The next section includes 3 chapters addressing “Evidence-based medicine: Defense and criticism,” “The alchemy of meta-analysis,” and “Bayesian statistics: Why your opinion counts.” The most interesting and practical is the chapter on meta-analysis. Interestingly, as Dr. Ghaemi points out, meta-analysis is the product of psychiatry and was developed to refute the criticism of Hans Eysenck that psychotherapies were ineffective. Meta-analysis can help clarify many things and provide some systematic way of putting together all data on a specific topic and interpreting them. Nevertheless, as the author points out, “meta-analysis is never more valid than an equally large single randomized controlled trial” (p 98). The chapter on Bayesian statistics is interesting and important, but also quite difficult to comprehend.
The final part is not so much about statistics as about some contemporary issues related to publishing and financing research. It consists of 4 chapters, “How journal articles get published,” “How scientific research impacts practice,” “Dollars, data, and drugs,” and “Bioethics and the clinician/researcher divide.” The chapter on how articles get published is mostly a criticism of the peer review process. The author suggests, among others, that “the most prestigious journals usually do not publish the most original or novel articles; this is because the peer review process is inherently conservative” (p 115). The following chapter is an astute criticism of the impact factor and intangibles of coauthorship of various large pharmaceutically funded trials. The criticism continues in the “Dollars, data, and drugs” chapter. The author discusses important issues such as ghost authorship, unpublished negative studies, and disease mongering. The final chapter focuses on some ethical issues of modern research and publishing. I think most readers would appreciate the summary statement of this chapter, “one cannot be a good clinician unless one is a good researcher, and one cannot be a good researcher unless one is a good clinician. Good clinical practice shares all the features of good research: careful observation, attention to bias and chance, replication, reasoned inference of causation” (p 130).
The book also contains an Appendix, “Regression models and multivariable analysis,” that provides some instruction on how to conduct regression analysis.
As the author points out in the Preface, “This book does not seek to teach you how to do statistics; it seeks to teach you how to understand statistics. It is for the clinician or researcher who wants to understand what he or she is doing or seeing; not for a statistician who wants to run a specific test” (p xii). The book mostly fulfills this goal (though, as I pointed out, the Bayesian statistics is difficult to understand in spite of all the author’s efforts). It is well written, straight to the point, thoughtful. The last part is mostly an opinionated view of various problems of contemporary science and scientific publishing, but it is mostly right on target. The book could be used as a complementary text (in addition to a text reviewing how to do some basic statistical analyses) to teach residents and others to understand statistics and its merits. Even busy clinicians may find this book useful reading, applicable to their reading of scientific literature.
Annals of Clinical Psychiatry ©2010 Quadrant HealthCom Inc.