Calibrated Agent Simulations for Combined Analysis of Drinking Etiologies (CASCADE)

 描述(由申请人提供)：众所周知，饮酒是导致各种健康状况和伤害的主要风险因素。该项目的重点是美国和英国的人群，据估计，每100人中就有3到4人死于酒精消费。增加疾病和伤害风险的饮酒模式是复杂的，但包括既往饮酒史、平均饮酒量以及一次喝五杯或更多饮料的频率。尽管酒精是一种有害物质，但在美国和英国的社会中，酒精的使用历史悠久，因此这些国家的立法者一直不愿收紧对向消费者销售酒精饮料的控制。特别是，在做出任何决定之前，政策制定者热衷于使用计算机模型来帮助理解未来任何政策可能产生的影响，比如人们如何改变饮酒方式以及随后对与酒精相关的伤害的影响。计算机模型帮助收集了一些关于特定饮酒模式原因的证据，以及干预(例如，增税)可能如何改变这些模式的证据。他们还成功地将这些消费变化与健康状况不佳、住院和过早死亡的变化联系起来。然而，这些模型也有一些弱点：它们没有涵盖导致人们饮酒的所有原因；它们在预测政策效果时，没有完全考虑到证据基础中的所有不确定性；它们的构建方式使得它们做出的预测不容易被检验。在这个项目中，我们的目标是改变用于酒精政策模型的方法，以克服所有这三个弱点。要做到这一点，我们需要将饮酒视为一个人的有意识的决定或习惯，但包含在一个更广泛的系统中，受到其他人、机构和法规的影响。我们的第一个具体目标是测试四种理论，本认为这些理论可以解释1979年至2015年期间五个人群(美国、三个州-加利福尼亚州、纽约州和德克萨斯州-以及英格兰)的饮酒行为。这些理论基于个性变化、家庭、朋友和父母的影响、人们成长的时代文化的持久影响，以及消费者的选择。我们的第二个具体目标是开发一个新的基于系统的模型，整合理论的有用部分，并测试其代表1979至2015年五个人群中饮酒模式的能力。我们将使用现代统计方法来确保模型的预测正确地考虑到不确定性。我们的第三个具体目标是扩展基于系统的模型，以便它们也可以产生与酒精相关的伤害估计，我们将使用酒精中毒和肝硬变作为结果来测试这些能力。我们的第四个具体目标是扩展该模型，以便它可以估计政策的影响，我们将使用它来评估税收和最低法定饮酒年龄法律的历史变化，并预测2016-2025年家庭医生对酒精滥用的最低定价和筛查。最后，我们将使项目中开发的计算机模型公开供所有人使用。

 DESCRIPTION (provided by applicant): Alcohol consumption is known to be a major risk factor for a wide range of health conditions and injuries. This project focuses on populations in the US and UK, where alcohol consumption is estimated to be responsible for between 3 and 4 deaths in every 100. The patterns of drinking that increase the risk of disease and injury are complex, but include prior history of drinking, average quantities consumed, and how often five or more drinks are consumed in a single occasion. Despite being a harmful substance, alcohol has a long history of use in both US and UK societies and so legislators in these countries have been reluctant to tighten controls on the sale of alcoholic beverages to consumers. In particular, before making any decisions, policymakers are keen to use computer models to help understand what the likely impact of any future policy is likely to be, in terms of how people change their drinking and the subsequent effects on alcohol-related harms. The computer models have helped to bring together some of the evidence on the causes of particular patterns of drinking, and how an intervention (for example, increasing taxes) might change these patterns. They have also been successful in relating these changes in consumption to changes in ill health, hospital admissions, and premature deaths. However the models also have some weaknesses: they do not cover all of the reasons that cause people to drink; they do not fully account for all of the uncertainty in the evidence base in their predictions of policy effect; and they are built in such a way that the predictions they make cannot easily be tested. In this project, our goal is to transform the methods used for alcohol policy models to overcome all three weaknesses. To do this, we need to consider drinking as being a conscious decision or habit of an individual person, but contained within a wider system of influences from other people, institutions, and regulations. Our first specific aim is to test four theories that have ben contended to explain drinking behaviors within five populations (the US as a whole, three states - California, New York and Texas - and also England) over the period 1979 to 2015. The theories are based on personality changes, influence of family, friends and parents, lasting effects of the culture of the era when people grew up, and consumer choice. Our second specific aim is to develop a new systems-based model, integrating useful parts of the theories, and test its ability to represent drinking patterns in the five populations over 1979 to 2015. We will use modern statistical methods to ensure that the models' predictions properly account for uncertainty. Our third specific aim is to extend the systems-based models so they also produce estimates of alcohol-related harms, and we will test these capabilities using alcohol poisoning and liver cirrhosis as outcomes. Our fourth specific aim is to extend the model so that it can estimate the impact of policies, and we will use it to evaluate historical changes in taxation and minimum legal drinking age laws and to make predictions over 2016-2025 for minimum pricing and screening at family doctors for alcohol misuse. Finally, we will make the computer models developed in the project publicly available for everyone to use.