Excessive alcohol use has many negative health consequences, including higher risk of mortality. In addition, the estimated economic cost of excessive drinking is more than 1% of the gross national product in high- income and middle-income countries (Rehm et al., 2009). The over-arching goal of this application is to improve the definition and assessment of alcohol use disorder (AUD). Our proposal is consistent with the goals outlined in the most recent NIAAA Strategic Plan (2009-2014). Most broadly, our study will develop and validate optimal diagnostic criteria sets and algorithms (ODCSAs) for the diagnosis of AUD that we anticipate will perform substantially better than existing diagnostic criteria (i.e. DSM-5 and ICD-10). We will employ emerging methods and develop and apply new methods in statistics and classification science to existing large population-based epidemiological data sets (the National Epidemiological Survey of Alcoholism and Related Conditions, NESARC; the National Longitudinal Alcohol Epidemiologic Survey, NLAES; the National Survey on Drug Use and Health, NSDUH), a large multi-site sample enriched for high density of AUDs (the Collaborative Study on the Genetics of Alcoholism, COGA), a major, multi-site treatment outcome study (Combining Medications and Behavioral Interventions; COMBINE) and two smaller non-clinical and clinical data sets spanning a decade or more of participants lives in order to refine assessment methods, diagnostic criteria, and diagnostic algorithms for AUDs. Specifically, by employing state-of-the art techniques in data harmonization and computational statistics, we will identify those criteria sets and decision rules that will optimize the associaton with heaviness of alcohol consumption, both within and across data sets. These new ODCSAs will then be subjected to extensive evaluation of their structural validity and external validity. n addition, we will conduct extensive validation analyses of DSM-5, ICD-10, and draft ICD-11 criteria, because these are of interest in their own right, and because that will allow us to benchmark our new ODCSAs with the current standards. We believe the development of empirically-based diagnosis for AUD will improve research on the causes and correlates of AUD and should lead to improved diagnostic criteria for clinical practice. The methods developed for this application should have broad applicability for developing diagnostic criteria or assessment instruments for a range of clinically important constructs when there are no clear gold standards. The proposed research exploits increased availability of data sharing, new developments in data harmonization, and advances in computational science.
Alcohol use disorder (AUD) represents a major public health problem and incur enormous costs to our society. However, existing approaches to diagnosis are most likely suboptimal. This proposal will improve the definition and assessment of alcohol use disorder (AUD) which, in turn, can facilitate research on the causes and correlates of alcohol-related problems, improve clinical practice, and lead to more meaningful public health statistics. More generally, the methodology developed in the proposal is seen as having broad applicability to a range of clinical and public health issues.
