Prevalence and Determinants of Alcohol Consumption among Schoolchildren in Barcelona, Spain
Villalbi, Joan R., Comin, Eva, Nebot, Manel, Murillo, Carles, Journal of School Health
Prevalence and Determinants of Alcohol Consumption Among Schoolchildren in Barcelona, Spain
Alcohol consumption has deep cultural roots in Western society, particularly in the wine-growing, Mediterranean region of Europe. It contributes to the development of several health problems in these countries, including chronic organic disorders, accidents, and effects on fetal and child development.  While effects from moderate alcohol consumption have been debated, consensus exists in Spain that alcohol abuse represents a major social and health problem. [2-6] Spain ranks second to France in international comparisons of alcohol consumption; 52% of adults drink daily, and mortality due to liver cirrhosis ranks as the sixth cause of death both in terms of death rates and in potential years of life lost. Consumption begins at an early age in wine-producing countries, with a high level of social acceptance for alcohol and fairly explicit social inducements toward its use. 
While previous studies have stressed characteristics associated with alcohol consumption by schoolchildren in Spain, [7,8] researchers know little about factors influencing earlier stages and experimentation. Within a broader study examining living habits related to health among studies in fifth and eighth grade, a survey was conducted of the prevalence, determining factors, and attitudes related to alcoholic beverage consumption in Barcelona, Spain. With 1.7 million inhabitants, Barcelona is a main economic, industrial, and cultural center and the capital of Catalonia (population 6 million), a culturally and socially distinct community within Spain that has received large migrations from rural areas in the sixties and seventies.
This survey included a representative sample of fifth and eighth grade students, approximately 10 and 13 years old, respectively. Variables studied included frequency and type of alcohol consumption, attitude, opinions, and beliefs of students regarding alcohol, money available to them, and perceptions of consumption by significant others (parents, friends, siblings), with sociodemographic data: age, gender, socio-economic status of neighborhood, and type of school. A questionnaires was adapted in the Catalan and Spanish languages following guidelines for self-administered questionnaires from standard models developed under auspices of the World Health Organization and used in other surveys studying tobacco and alcohol consumption in schoolchildren. [10,11] Data were collected between January and May 1987.
A randomly selected cluster sample was obtained and stratified according to three variables; grade level (fifth and eighth grades), type of school (public, private with global or partial public subsidies, and private for profit), and size of class. Class was used as the sampling unit. Stratification partially compensated for the increase in variability that may accompany this type of sampling.  Therefore, it seemed reasonable to expect that both grade (through age) and type of school (public or private, related to social class) are associated with alcohol consumption. To estimate questionnaire reliability, replication of prevalence data obtained previously was analyzed using Kendall's tau  in four classrooms, where the survey was repeated four to eight weeks after the initial questionnaire. Kendall's tau values ranked between 0.54 and 0.75 for repeatability of alcohol consumption and having gotten drunk.
In addition to stratified analysis by grade and gender, using chi-square tests for comparison of proportions, discriminant analysis was used to study the overall influence of variables associated with alcohol consumption. [14,15] An operational definition of excessive or problem alcohol consumption was prepared for this purpose, given that occasional social use is the norm in the environment. This procedure required creating several models that differentiate among patterns of use as the dependent variable. …