Pancreatitis mortality and population level alcohol consumption: taking the science a step forward

Mats Ramstedt's paper on pancreatitis mortality time-series covering 14 countries adds a new mortality classification to the growing body of research on the population-level relationships between alcohol consumption measures and mortality causes (Ramstedt 2004). Death from acute or chronic pancreatitis occurs relatively rarely, making their combination (as well as the pooling of rates for men and women) necessary for this analysis, but also making individual-level prospective studies difficult. Aggregate analyses utilize alcoholic beverage sales data, which is more comprehensive and accurately measured than self-reported consumption data offering an alternative yet complementary perspective on alcohol consumption's role in the etiology of pancreatitis. Perhaps the most important contribution of aggregate analyses is the use of comparable models across countries and causes in identifying differences in the magnitude of effects, and indeed the existence of an effect at all, to illuminate elements of underlying relationships. While changes in per capita consumption of alcohol have been generally found to shift the entire consumption distribution (Skog 1985), they can be related to (or mask) underlying changes in the structure of alcohol consumption. Changes in consumption may be related to policy adjustments, economic cycles and economic growth, tourism and cross-border sales, demographic shifts in the age, education or ethnic structure of the population and through birth cohort differences drinking patterns. Mortality rates are also influenced by other types of mortality that may compete for the lives of heavy drinkers, such as cirrhosis, heart disease, accidents and many others. Given the number of countries and the long time period utilized in these analyses, it is likely that many or all of these factors are involved in consumption and mortality trends. As such, considerable heterogeneity of results would be expected and, in fact, is found. While it seems clear that the relationship is confirmed, its estimated size and strength may reflect more about these underlying factors than about attributable fractions or the average impact of a litre of consumption. For example, birth cohort shifts in drinking patterns and levels may play an important role in alcohol trends (Kerr et al. 2004), and this type of change could be reflected in the weak relationships seen in Southern Europe and Canada. Ramstedt is careful to consider alternate models of the lag structure and linearity of the relationship, as these are not theoretically determined for all cases, and may in fact differ based on the underlying relationships. Key areas for comparative extension of these models could utilize beverage-specific consumption, alternative pooling assumptions and, particularly, they could include factors known to interact with alcohol in the etiology of pancreatitis. Morton et al. (2004) followed over 100 000 individuals, some of whom later developed pancreatitis. They found that smokers were at increased risk of alcohol-related pancreatitis and that coffee consumption reduced this risk. Data are available on both smoking and coffee consumption in these countries and these may explain differences in the relationship across counties and could modify the relationship found within each country, as changes have also occurred over time. Beverage-specific models are possible for most counties and separating consumption in this way can sometimes identify particular drinking patterns, demographic groups or birth cohorts that are associated more closely with a beverage. Pooling across countries is a difficult issue, given the clear heterogeneity of relationships. The cautious approach adopted here and in the ECAS (Norström & Skog 2001) and Canadian (Ramstedt 2003) time-series analyses is reasonable, and comparable to previous analyses of other alcohol-related mortality causes. However, it does not take advantage of the possibility that jointly estimated panel models could cut through influence of unobserved confounders, as argued by Baltagi & Griffin (1995), in their analysis of spirits consumption and taxation in the United States. It is possible that this type of approach (including smoking and coffee consumption) would result in a single answer as to effect of alcohol on pancreatitis. Nevertheless, this paper represents an important step toward a pool of global comparative analyses that will greatly aid our understanding of the diversity of alcohol consumption behaviors and their roles in the many mortality causes linked to alcohol.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.