Deciphering Global Epidemics: Analytical Approaches to the Disease Records of World Cities, 1888-1912 (review)

Andrew Cliff, Peter Haggett, and Matthew Smallman-Raynor. Deciphering Global Epidemics: Analytical Approaches to the Disease Records of World Cities, 1888-1912. Cambridge Studies in Historical Geography, no. 26. Cambridge: Cambridge University Press, 1998. xxiii + 469 pp. Ill. $74.95 (cloth), $29.95 (paperbound).

In 1878, Surgeon-General John M. Woodworth initiated a new system to protect native shores against disease carried by immigrants and foreign commerce: U.S. consuls around the world were to report weekly on local disease outbreaks, information to be supplemented with domestic reports. Woodworth's Weekly Abstract of Sanitary Reports, the grandfather of today's Morbidity and Mortality Weekly Report, assembled vast amounts of data on the worldwide incidence of contagious disease. More than one hundred years later, an innovative team of medical geographers has put Woodworth's data to use in analyzing patterns of infectious disease in the era of the epidemiologic transition.

From the Weekly Abstract, the authors have assembled a database tracking six diseases in one hundred cities distributed in ten regions for the period 1888-1912. The diseases are diphtheria, measles, scarlet fever, whooping cough, tuberculosis, and "enteric fever" (which the authors take to be typhoid and paratyphoid). Mortality from these causes is then analyzed in three chapters that form the analytical core of the book, looking respectively at mortality changes worldwide, intra- and interregionally, and in a select group of individual cities: London, New York, Paris, Vienna, St. Petersburg, Chicago, Philadelphia, Moscow, and Bombay. As the authors argue, the breadth of their data gives them a perspective that city or even countrywide studies cannot provide. A concluding chapter briefly traces the subsequent twentieth-century experience with these six diseases, drawing on national-level mortality data after 1912 when the Weekly Abstract dropped its international coverage.

For some time now, the authors have been working at the cutting edge of medical geography, analyzing the ways in which diseases are distributed in time and space.¹ Their interests lie both in developing new analytical methods for "spatial epidemiology" and in demonstrating the value of such historical geography to contemporary infectious disease control. Toward the first end, they rely here on methods derived from econometrics (Box-Jenkins time series analysis), psychology (principal components analysis), and physics (spectral analysis), among others. Few but economic historians will be able to follow these analyses, much less evaluate them judiciously. In many respects, the book is a model of social science, with at least one-third of its pages devoted to a critical discussion of sources and statistical methods. For historians, however, the yield is disappointing: some intriguing epidemiologic facts, but little in the way of sustained historical analysis.

Thus, it is useful to know that worldwide mortality from diphtheria and enteric fever declined absolutely over this period; that scarlet fever mortality declined relative to other causes; and that tuberculosis was on the increase. It is useful also to know that smaller cities generally had higher mortality rates than larger cities, especially in the "developing" regions of the world (as distinct from Northern and Western Europe, North America, and the British Isles). But we get little focused attention to why these patterns should have occurred, or why, for example, rises in mortality should have followed a hierarchical model (from larger to smaller cities) rather than one of propinquity (among cities that are physically closer to one another, regardless of their size). There are geographical and historical theories about why diseases, like manufactured goods, travel from larger places to smaller ones, but they go unexplored here.

Part of the problem lies in the authors' decision to wring out the maximal information from their data set: virtually all of their effort goes to ingenious methods for examining the cyclical nature of epidemics or the similarity/dissimilarity of multiple time series, and little energy is left for...