Alcoholic subtypes based on multiple assessment domains. Validation against treatment outcome

Research in the area of subtype development within alcoholic populations has undergone a number of changes over the recent years. First, there has been a shift from subtype derivation based on a priori comparative approaches toward a more recent use of a posteriori correlational approaches such as cluster analysis. Second, there has been an increased multivariate focus in which multiple measures from a number of practically and/or theoretically relevant assessment domains are entered into the classificatory process. Third, there has been an increased emphasis on the external validation of derived clusters. The present chapter briefly reviews these issues, using examples of research from the Seattle Veterans Administration Medical Center to illustrate each of these shifts across a series of studies focusing on the development of alcoholic subtypes. Finally, the results of a recent subtype identification and validation study will be presented to illustrate the difficulties in and advantages of a multidomain cluster analytic strategy.