British Journal of Addiction (1991) 86, 1203-1210 THE COLLABORATIVE ALCOHOL-RELATED LONGITUDINAL PROJECT Preliminary results from a meta-analysis of drinking behavior in multiple longitudinal studies'^ KAYE MIDDLETON FILLMORE, ELIZABETH HARTKA, BRYAN M. JOHNSTONE, E. VICTOR LEINO, MICHELLE MOTOYOSHI & MARK T. TEMPLE Institute for Health & Aging, Department of Social and Behavioral Sciences, University of California, San Francisco, 201 Filbert Street, Suite 500, San Francisco, California 94133-3203, USA With the help of the following collaborators: Salme Ahlstrom (Finland), Peter Allebeck (Sweden), Arvid Amundsen (Norway), Jules Angst (Switzerland), Gellisse Bagnall (Scotland), Ann Brunswick (USA), Remi Cadoret (USA), Sally Casswell (New Zealand), Lorraine de Labry (USA), Norman Giesbrecht (Canada), Bridget Gram (USA), Thomas Greenfield (USA), Joel Grube (USA), Bernd Guether (Federal Republic of Germany), Thomas Harford (USA), Ludek Kubicka (Czechoslovakia), Steve Manske (Canada), Mark Morgan (Ireland), Harold Mulford (USA), Leif Ojesjo (Sweden), David Peck (Scotland), Martin Plant (Scotland), Chris Power (UK), Lee Robins (USA), Anders Romelsjo (Sweden), David Rosen (USA & Shetland Islands), Ronald Schlegel (Canada), Martin Sieber (Switzerland), Soren Sigvardsson (Sweden), Rainer Silbereisen (Federal Republic of Germany), Ronald Stall (USA), Meir Teichman (Israel), Richard Wilsnack (USA), and Sharon Wilsnack (USA) and consultants: Don Cahalan (USA), Marcus Grant (WHO), Larry Hedges (USA), Harold Holder (USA), Andrea Mitchell (USA), Leonard Pearlin (USA), Robert Straus (USA), Leland Towie (USA) and Mark Wilson (USA). Abstract This paper introduces the scope and rationale of The Collaborative Alcohol-Related Longitudinal Project and describes the individual longitudinal studies which contribute to this large collaborative project, representing studies from 15 countries, ft also serves as an introduction to four reports of the preliminary findings from the project. The project is distinguished by (I) its interdisciplinary research approach which has assembled a multidisciplinary group of scholars to direct and interpret analyses, (2) its use of primary data from multiple longitudinal studies, (3) the parallel analyses of primary data from multiple studies, using comparable measures across studies recoded to a standard format and common analytic model, and (4) its use of meta- analysis to combine results across studies. Its research objectives include determining the cross-study consistency of findings of (1) the incidence and chronicity of drinking patterns and problems, (2) exogeneous factors which initiate and alter drinking careers, (3) socio-behavioral factors measured in childhood and adolescence which predict adult drinking problems, (4) inter-generational biological and social factors which predict adult drinking problems, and (5) aggregate-level factors which account for study differences. The method of sampling of studies from the world's alcohol-related general population longitudinal research is described. *This work was supponed by a National Institute on Alcohol Abuse and Alcoholism (NIAAA) gram (#R01 AAO7O34) and by a NIAAA Research Scientist Development Award (#K0! AAOOO73) to the first author. The Collaborative Project is included in the plan of work of NIAAA as a World Health Organization (WHO) Collaborating Center on Research and Training in Alcohol-Related Problems, and is aiso affiliated with the WHO Global Program on Prevention and Control of Alcohol and Drug Abuse. Order of authorship in the Collaborative Alcohol-Related Longitudinal Project is designated by the following criteria; (a) the first author has taken principal responsibility for organizing and writing the research paper; (b) persons making substantial contributions foUow the first author in alphabetical order; and (c) collaborators, having reviewed the paper and its findings in accordance with accuracy and representation of their data and project goals. 1203 1204 K. Middleton et al. Introduction recoded into a standard format, to enable the This collection of papers gives the first reports testing of hypotheses and assumptions of from a group of scholars from 15 countries, alternative disciplines; representing 39 longitudinal data sets—together (3) uses data sets of longitudinal measurement. known as The Collaborative Alcohol-Related Longitudinal Project. The scope and rationale of The project's objectives are governed by a group the project are briefly described in this introduc- of coUborators and consultants. The direct involve- tory paper. Following this overview are four pa- ment of scholars responsible for collecting the data pers. The first describes the design of the project and analysis of the individual studies ensures that (Johnstooe et al., 1991) while the remaining three specific information from each study is accurate in present preliminary research findings: (a) indivi- the comparative analyses, and informs the interpre- dual-level and group-level description of life tation of results with knowledge of specific cultural course variation for two drinking variables (fre- and historical contexts of the studies. The scholars quency of drinking per month and quantity per differ with respect to disciplinary orientation, en- typical occasion), both cross-sectionally and longi- hancing assessment of research objectives and tudinally (Fillmore et al., 1991), (b) prediction of dialogue about them. quantity per typical occasion over time by indivi- Studies included in the project were initially dual-level changes in marital status and employ- chosen on the basis of a review of the world's ment status within specified age/sex groups alcohol-related longitudinal literature (Fillmore, (Temple et al., 1991), and (c) prediction of 1988). General population longitudinal studies and individual-level depressive symptomatology from adoptee studies were the targets of sampling. individual-level quantity per typical occasion and Consideration was given to the scope of alcohol- vice-versa (Hartka et al., 1991). related questions from each study's published ac- counts, sampling methodology, and sample size, narrowing the final pool to 39 studies. Principal investigators of 36 general population longitudinal Scope and rationale of the collaborative studies and three adoptee studies were contacted to project inquire as to their interest in participating in a The project assesses the extent to which cross-study collaborative study and filled out questionnaires longitudinal findings from different cultural and regarding items in their study's questionnaires, temporal contexts replicate. This question has sampling methodology and availability of data. previously been sorely neglected, thereby limiting Four general population studies were excluded generalization of findings. Although longitudinal after the first author evaluated questionnaire re- measurement is perhaps one of the strongest re- sponses (reasons for exclusion: original data were search strategies for describing and predicting lost, there was no Time 2 follow-up planned and, drinking patterns and problems across the life in two cases, the measures or methodologies were course, and almost 100 such general population judged to be incompatible with the research ques- studies have been initiated since the 1920s, the tions the larger project would address). Four of the extent to which results from such studies can be remaining 32 in the general population pool, and 1 replicated across age cohorts, historical periods, and out of 3 from the adoptee study pool either did not national setting in the description and prediction of respond to the inquiry or refused to participate. drinking patterns and problems has not been Since the initiation of the project, 9 general systematically examined (Fillmore, 1988). This population longitudinal studies (3 of them from the collaborative study is a first step in addressing this same institute) have been added to the archive. problem. While this sample of studies is generally represen- The Collaborative Project: tative of alcohol-related longitudinal studies exist- ing in the world today, it is clearly a biased group (1) represents interdisciplinary research, having with respect to historical period (the majority of assembled a multi-disciplinary group of the studies were initiated in the last 30 years) and scholars to direct and interpret analyses; cultural context (studies from Northern Europe (2) utilizes primary data from multiple longitu- and North America are over-represented; there are dinal data sets (the majority being of the no studies included from the developing world or general population), assembled at the from Southern Europe). i ' University of California, San Francisco, and Meta-analysis of drinking behavior 1205 Table 1. Individual studies contributing to the Collaborative Project Alcohol-Related Longitudinal Project, study code, key personnel and source of funding Key personnel (collaborator Study (code in italics) with this project underlined) Source of support A Prospective Longitudinal Study of M. Bohman, C. R. Cloninger, S. Swedish Medical Research Council Adoptees (Sweden-s) Sigvardsson (University oF Adoption USA Public Health Service Umea), A. Von Knorring Alcohol & Drug Use in Young Adults S. Manske & R. Schlegel National Health Research & Develop- {Canada-m) (University of Waterloo) ment Program (6606-2994-42) & So- cial Sciences & Humanities Council of Canada (410-85-1176) Alcohol & Psychoactive Drug Use Z. Bamea, G. Rahav, M. Teich-