Karl Mann1, Stephanie S O'Malley2, Ralitza Gueorguieva3, Ran Wu2, Patrick G O'Connor4, Constance Weisner5,6, Lisa M Fucito2, Sabine Hoffmann1. 1. Department of Addictive Behavior and Addiction Medicine, Central Institute of Mental Health, Medical Faculty Mannheim, Heidelberg University, 68159, Germany. 2. Department of Psychiatry, Yale University School of Medicine, New Haven, CT 06519, USA. 3. Yale University School of Public Health and School of Medicine, New Haven, CT 06520, USA. 4. Department of Internal Medicine, Yale University School of Medicine, New Haven, CT 06519, USA. 5. Kaiser Permanente Division of Research, Oakland, CA 94612, USA. 6. Department of Psychiatry, University of California at San Francisco, San Francisco, CA 94143, USA.
Abstract
BACKGROUND: The goal of the current study was to use tree-based methods (Zhang and Singer, 2010, Recursive Partitioning and Applications, 2nd ed. Springer, New York) to identify predictors of abstinence from heavy drinking in COMBINE (Anton et al. JAMA 2006; 295:2003), the largest study of pharmacotherapy for alcoholism in the United States to date, and to validate these results in PREDICT (Mann et al. Addict Biol 2012; 18:937), a parallel study conducted in Germany. METHODS: We compared a classification tree constructed according to purely statistical criteria to a tree constructed according to a combination of statistical criteria and clinical considerations for prediction of no heavy drinking during treatment in COMBINE. We considered over 100 baseline predictors. The tree approach was compared to logistic regression. The trees and a deterministic forest identified the most important predictors of no heavy drinking for direct testing in PREDICT. RESULTS: The tree built using both clinical and statistical considerations consisted of 4 splits based on consecutive days of abstinence (CDA) prior to randomization, age, family history of alcoholism, and confidence to resist drinking in response to withdrawal and urges. The tree based on statistical considerations with 4 splits also split on CDA and age but also on gamma-glutamyl transferase level and drinking goal. Deterministic forest identified CDA, age, and drinking goal as the most important predictors. Backward elimination logistic regression among the top 18 predictors identified in the deterministic forest analyses identified only age and CDA as significant main effects. Longer CDA and goal of complete abstinence were associated with better outcomes in both data sets. CONCLUSIONS: The most reliable predictors of abstinence from heavy drinking were CDA and drinking goal. Trees provide binary decision rules and straightforward graphical representations for identification of subgroups based on response and may be easier to implement in clinical settings.
BACKGROUND: The goal of the current study was to use tree-based methods (Zhang and Singer, 2010, Recursive Partitioning and Applications, 2nd ed. Springer, New York) to identify predictors of abstinence from heavy drinking in COMBINE (Anton et al. JAMA 2006; 295:2003), the largest study of pharmacotherapy for alcoholism in the United States to date, and to validate these results in PREDICT (Mann et al. Addict Biol 2012; 18:937), a parallel study conducted in Germany. METHODS: We compared a classification tree constructed according to purely statistical criteria to a tree constructed according to a combination of statistical criteria and clinical considerations for prediction of no heavy drinking during treatment in COMBINE. We considered over 100 baseline predictors. The tree approach was compared to logistic regression. The trees and a deterministic forest identified the most important predictors of no heavy drinking for direct testing in PREDICT. RESULTS: The tree built using both clinical and statistical considerations consisted of 4 splits based on consecutive days of abstinence (CDA) prior to randomization, age, family history of alcoholism, and confidence to resist drinking in response to withdrawal and urges. The tree based on statistical considerations with 4 splits also split on CDA and age but also on gamma-glutamyl transferase level and drinking goal. Deterministic forest identified CDA, age, and drinking goal as the most important predictors. Backward elimination logistic regression among the top 18 predictors identified in the deterministic forest analyses identified only age and CDA as significant main effects. Longer CDA and goal of complete abstinence were associated with better outcomes in both data sets. CONCLUSIONS: The most reliable predictors of abstinence from heavy drinking were CDA and drinking goal. Trees provide binary decision rules and straightforward graphical representations for identification of subgroups based on response and may be easier to implement in clinical settings.
Authors: Aleksandra E Zgierska; Cindy A Burzinski; Marlon P Mundt; Andrew S McClintock; Jennifer Cox; Christopher L Coe; Michael M Miller; Michael F Fleming Journal: J Subst Abuse Treat Date: 2019-01-17
Authors: Ralitza Gueorguieva; Ran Wu; Wan-Min Tsai; Patrick G O'Connor; Lisa Fucito; Heping Zhang; Stephanie S O'Malley Journal: Eur Neuropsychopharmacol Date: 2015-06-20 Impact factor: 4.600
Authors: Kelly E Dunn; Joseph A Harrison; Jeannie-Marie Leoutsakos; Dingfen Han; Eric C Strain Journal: Alcohol Alcohol Date: 2016-08-27 Impact factor: 2.826
Authors: Gonzalo Salazar de Pablo; Erich Studerus; Julio Vaquerizo-Serrano; Jessica Irving; Ana Catalan; Dominic Oliver; Helen Baldwin; Andrea Danese; Seena Fazel; Ewout W Steyerberg; Daniel Stahl; Paolo Fusar-Poli Journal: Schizophr Bull Date: 2021-03-16 Impact factor: 9.306
Authors: Joanna M Biernacka; Brandon J Coombes; Anthony Batzler; Ada Man-Choi Ho; Jennifer R Geske; Josef Frank; Colin Hodgkinson; Michelle Skime; Colin Colby; Lea Zillich; Sofia Pozsonyiova; Ming-Fen Ho; Falk Kiefer; Marcella Rietschel; Richard Weinshilboum; Stephanie S O'Malley; Karl Mann; Ray Anton; David Goldman; Victor M Karpyak Journal: Neuropsychopharmacology Date: 2021-07-23 Impact factor: 7.853