This revised application responds to PA-05-088 for secondary analysis of existing data to enhance understanding of drinking patterns and the epidemiology of alcohol-related problems-here, mortality. The PA's objectives will be met by aims addressing risk and protective factors in the US population and in white, black and Hispanic subpopulations of both genders (helping elucidate disparities in alcohol-related mortality), impact of drinking history on mortality, and methodological issues in pattern measurement and analysis including propensity score strategies. Recent alcohol mortality studies, including a successful R01 (AA10960, T.K. Greenfield, Principal Investigator), have found preliminary evidence of the role of drinking pattern, i.e., heavy drinking among low to mid-volume drinkers, in risk of all-cause mortality. Earlier analyses have been limited by lack of power, inadequate measurement, or both. This application uses two waves of the National Alcohol Survey (NAS) with unusually thorough measurement of drinking pattern, lifetime heavy drinking, and alcohol-related problems. It will extend from 11 to 21 (or if possible 22) years the National Death Index (NDI) follow up of the 1984 NAS (n = 5,221, response rate of 74 percent) adding a new 11- or 12-year follow up of the 1995 NAS (n = 4,950, response rate = 79 percent). Both surveys (ages 18 and older) were conducted face-to-face by Temple University ISR, used probability sampling with large black and Hispanic over-samples, and assessed many relevant identical items and scales. Drinking pattern measures include 12-month quantity/frequency of beer, wine, liquor and overall drinking, most drinks/day, former drinking and largest quantity, ever 5+ weekly, age of onset, etc. Smoking, drug use, depression/ demoralization, and many other risk factors are assessed. The common design permits pooling NASs, more than tripling number of expected deaths (to over 1,800 compared to the earlier R01's 532 deaths), yielding considerably greater power than heretofore for testing hypotheses about all-cause, ischemic heart disease (IHD) and cancer mortality, with exploratory analyses on injury and liver cirrhosis mortality. Analyses will disaggregate by gender and, where possible, by this and ethnic/racial group. Interactions between alcohol consumption and ethnicity will be modeled; theoretically important covariates and their interactions with alcohol measures will be added in a series of planned analyses, e.g., CES-D depression, social isolation, and other risk factors such as history of heavy drinking and alcohol problems, quitting because of health problems, alcohol treatment, drug use, and other variables. Results will inform public health strategies, alcohol policy development, and help disentangle pattern effects on mortality. ? ? ? ?
