Changing validity of the cirrhosis mortality--alcoholic beverage sales construct: U.S. trends, 1970-1977

Data on cirrhosis mortality and alcoholic beverage sales were studied for the period 1970-1977 in the U.S. The relationship between cirrhosis mortality and sales was examined by sex and by type of beverage (beer, wine or distilled spirits). Time-series analyses showed that, during this period, total cirrhosis mortality decreased an average of 1.5% per year, sales increased 0.7% per year and alcohol-related cirrhosis mortality increased 2.6% per year. In any given year, however, these factors were highly correlated. Also, the relative ranking of the states remained stable, suggesting internal consistency of record-keeping. Misestimates of cirrhosis mortality based on sales were examined. Cross-sectional data indicated that sales were positively related to cirrhosis mortality but time-series analyses suggested that increases in sales were associated with decreases in total cirrhosis mortality and increases in the percentage of alcohol-related cirrhosis mortality. Time-series and regression analyses suggested the strong influence of factors external to the basic sales--cirrhosis mortality model and suggested that multiple indicators would be better than cirrhosis mortality and sales, alone or in combination, to estimate the magnitude of alcohol problems.