Why review a small dictionary on the market long enough to get to a third edition? There is ample reason. The first edition came out in 1983. "Epidemiology" is no longer only a tag for the field epidemiology once was: the study of disease outbreaks, prevalences of chronic diseases, and other questions about populations. Its then-small sibling, "clinical epidemiology," has grown well past adolescence and talks its own talk. And there is the not-yet tersely named discipline of "clinical trials research," which some clinical epidemiologists claim as part of the family. So this dictionary had to push out its boundaries to cover all of the terms that can be considered epidemiologic.

But surely a general medical dictionary, such as Stedman's Medical Dictionary, is adequate for these terms? The answer is yes and no. Many terms defined in the Last dictionary also appear in the latest edition of Stedman's, but many do not. For example, "gray literature," "queueing theory," and "utility-based units" are all excluded. A further shortcoming of general medical dictionaries is that their ways of consolidating otherwise scattered multiword terms into groups determined by shared generic terms, such as "curve," "disease," and "keratitis," can lead the reader in circles. For example, "receiver operating characteristic curve" appears in both the Last dictionary and in Stedman's. But if you go to "receiver" in Stedman's, all that you see is, "In chemistry, a vessel attached to a condenser to receive the product of distillation." If you think to go to "ROC," you find a cross-reference to "ROC curve" (and if you knew the acronym ROC, you probably did not need its definition). If you go to "curve," you do finally find "receiver operating characteristic curve." Admittedly, a frequent user of Stedman's would probably have gone directly to "curve" and run down the alphabetic list to "receiver."

The Last dictionary has other advantages. Its definitions often include much more detail. For example, the entry for "population attributable risk" includes formulas for the calculation of that risk. Many entries include references to relevant literature. Types of graphs are illustrated by examples.

The Last dictionary does have weaknesses. The worst of these is that the entry terms are set in small capitals (and initial capital letters are used for proper nouns, as in "INGELFINGER RULE") rather than in lowercase letters. The entry terms for multiterm abbreviations (for example, NRR for "net reproduction rate") are also set in small capitals. This ill-considered choice of small capitals for words that are properly set in lowercase letters (for example, "rule" rather than "RULE") and for multiterm abbreviations ("NNR" instead of "NNR") obscures the proper rule for setting such abbreviations in capital letters. Thus, the entry for APACHE appears as APACHE, which implies that a legitimate form would be "apache." (Incidentally, multiterm abbreviations are referred to as "acronyms," a term properly reserved for coinages such as "laser" that have become single words properly set in lowercase letters.) Some style flaws are conspicuous, such as the legalism "and/or" and the closing of sentences with "etc." Given the Oxford University Press's fastidious publication style in major reference works (for example, its general English dictionaries), it is surprising that the Last dictionary is flawed in such details. In coverage: Many synonyms are given, but some that one would expect to find are not. "Gray literature" is defined but without the synonymous term that is the standard formal term among librarians, "ephemeral literature." No comment is made on the frequent misuse in clinical literature of "incidence" for what should be accurately termed "cumulative occurrence."

Such weaknesses do not cripple the overall value of this specialized work. Altogether, it is a book that should be on the reference shelves of epidemiologists, both classic and clinical, and in the offices of journals that publish what these epidemiologists write. At its price, this hardbound book with content reflecting hours and hours of intellectual work by experts is a bargain. The International Epidemiological Association brings itself great credit for sponsoring it. I hope, however, that Oxford University Press will attend more closely in future editions to details of publication style.

Edward J. Huth, MD

University of Pennsylvania

Philadelphia, PA 19104