摘要：Objectives. We estimated the prevalence of any drinking and binge drinking from 2002 to 2012 and heavy drinking from 2005 to 2012 in every US county. Methods. We applied small area models to Behavioral Risk Factor Surveillance System data. These models incorporated spatial and temporal smoothing and explicitly accounted for methodological changes to the Behavioral Risk Factor Surveillance System during this period. Results. We found large differences between counties in all measures of alcohol use: in 2012, any drinking prevalence ranged from 11.0% to 78.7%, heavy drinking prevalence ranged from 2.4% to 22.4%, and binge drinking prevalence ranged from 5.9% to 36.0%. Moreover, there was wide variation in the proportion of all drinkers who engaged in heavy or binge drinking. Heavy and binge drinking prevalence increased in most counties between 2005 and 2012, but the magnitude of change varied considerably. Conclusions. There are large differences within the United States in levels and recent trends in alcohol use. These estimates should be used as an aid in designing and implementing targeted interventions and to monitor progress toward reducing the burden of excessive alcohol use. Excessive alcohol consumption is associated with many adverse health outcomes, including cancer, heart disease, stroke, liver cirrhosis, preterm birth, fetal alcohol syndrome, and unintentional and intentional injuries. 1–7 In 2010, approximately 88 600 deaths in the United States were attributable to alcohol, and the cost of excessive drinking has been estimated to exceed $220 billion per year. 8,9 In the United States, several ongoing surveys collect information on alcohol use. Most of these surveys are designed to produce national-level estimates only, whereas the Behavioral Risk Factor Surveillance System (BRFSS) and the National Survey on Drug Use and Health are designed to produce state-level estimates. More local estimates would be useful for identifying high-risk populations and for policymaking. Below the state level, however, information on alcohol use is limited: existing estimates are generally for select metropolitan areas only 10,11 or are derived by pooling data across a large number of years, making it difficult to assess trends. 12 We estimated trends in alcohol use at the county level. Specifically, we assessed the prevalence of any drinking (at least 1 drink of any alcoholic beverage in the past 30 days) and 2 measures of excessive alcohol use: the prevalence of heavy drinking (consuming, on average, more than 1 drink per day for women or 2 drinks per day for men in the past 30 days) and the prevalence of binge drinking (consuming at least 4 drinks for women or 5 drinks for men on a single occasion at least once in the past 30 days). We included both heavy and binge drinking because these 2 measures of excessive alcohol consumption are related to different health outcomes: binge drinking tends to increase the risk of short-term health effects (e.g., injuries 6 ), whereas heavy drinking tends to increase the risk of long-term health effects (e.g., cancers, 7 liver cirrhosis 5 ). Further, whereas most people who are heavy or binge drinkers according to these definitions do not suffer from an alcohol use disorder or alcohol dependence, these behaviors are associated with an elevated risk of developing these conditions. 13,14 We used small area models to produce annual estimates of any drinking, heavy drinking, and binge drinking prevalence at the county level on the basis of BRFSS data. These models incorporate a series of corrections to account for changes in questionnaire items related to alcohol consumption and for methodological changes regarding cell phones. We also combined these measures to estimate the prevalence of heavy or binge drinking among those who use any alcohol.
