Download this article now for $15.00.
Wartime Paris, Cirrhosis Mortality, and the Ceteris Paribus Assumption
Kaye Middleton Fillmore, Ron Roizen, Michael Farrell, William Kerr, Paul Lemmens
Objective: This article critiques the ceteris paribus assumption, which tacitly sustains the epidemiologic literature’s inference that the sharp decline in cirrhosis mortality observed in Paris during the Second World War derived from a sharp constriction in wine consumption. Paris’s wartime circumstances deviate substantially from the “all else being equal” assumption, and at least three other hypotheses for the cirrhosis decline may be contemplated. Method: Historical and statistical review. Results: Wartime Paris underwent tumultuous changes. Wine consumption did decline, but there were, as well, a myriad of other changes in diet and life experience, many involving new or heightened hardships, nutritional, experiential, institutional, health and mortality risks. Three competing hypotheses are presented: (1) A fraction of the candidates for cirrhosis mortality may have fallen to more sudden forms of death; (2) alcoholics, heavy drinkers and Paris’s clochard subpopu- LIVER CIRRHOSIS mortality rates plunged in Paris over the course of World War II. A mere 178 cirrhosis deaths were reported by 1945, down from an average of about 1,000 cirrhosis deaths per year over 3 prewar years, 1936- 1938. We revisit this remarkable epidemiological event because, over the course of the intervening decades, the Paris cirrhosis decline has come to occupy a not inconsiderable place in alcohol studies as an oft-cited and dramatic datum in the epidemiological literature on alcohol’s cirrhogenic responsibility. Wine was strictly rationed during 1942-1947 and was in short supply in wartime Paris. Thus, the sharp decline in the city’s cirrhosis rate may have increasingly come to be regarded as offering solid historical evidence of alcohol’s cirrhogenic responsibility in the present time, when longer-term associations between consumption and cirrhosis have weakened. The data originally illustrating this event were drawn from the Annuaire Statistique de la Ville de Paris, Année 1957 (Préfecture de la Seine, 1957), and were reproduced by Ledermann in 1964 (p. 357). Ledermann’s table contained the rates of all-cause mortality between 1912-1956 and cirrhosis deaths between 1907-1956 for the city; the data are also attributed to Dr. G. Pequignot. A footnote is attached to the relevant years of wine rationing: “Rationnement du vin: 1/2 litre à 1 litre par semaine.” The data from cirrhosis deaths have been transformed into a figure with which most contemporary alcohol researchers are quite familiar (Figure 1). The Paris curve, as it became known, appears in important discussions of alcohol’s cirrhogenic potentials (Bruun et al., 1975; Edwards et al., 1994; Lelbach, 1968; Skog, 1984; Terris, 1967). However, there has been a paucity of critical attention to it in the alcohol-epidemiological literature. Indeed, the figure itself is ordinarily presented with little or no caution at all, as if its epidemiologic message and validity were so clear-cut and straightforward as to speak for themselves. Numerous caveats and reservations might be attached to the famous Paris curve, however: some obvious, some more subtle. Such a critique may begin with the standard scientific presumption that any evidence derived from a “natural experiment” historical trend relies on the assumption that the “all else remained equal,” or the ceteris paribus, requirement was reasonably well met with lation may have been differentially likely to become removed from the city’s wartime population, whether by self-initiated departure, arrest and deportation, or death from other causes, even murder; and (3) there was mismeasurement in the cirrhosis mortality decline. Conclusions: The alcohol- cirrhosis connection provided the template for the alcohol research effort (now more than 20 years old) aimed at re-establishing scientific recognition of alcohol’s direct alcohol-problems-generating associations and causal responsibilities. In a time given to reports of weaker associations of the alcohol-cirrhosis connection, the place and importance of the Paris curve in the wider literature, as regards that connection, remains. For this reason, the Paris findings should be subjected to as much research scrutiny as they undoubtedly deserve. (J. Stud. Alcohol 63: 436- 446, 2002)
