[Discriminant analytic study of factors in relation to alcohol consumption of females]

On behalf of the Swiss Federal Commission against Alcoholism, a pilot study on drinking habits in women was made, in accordance with an earlier investigation of drinking habits in young men (Battegay et al. 1977). Employed women of Swiss nationality, in an industrial area in a northwestern agglomeration were questioned using a standardized questionnaire about appropriate facts. As participation in this inquiry was voluntary, only 436 (31%) of the 1362 employed women took part. Since, by the usual statistical procedures significant differences between heavier and lighter consumers were found, the extent of separation of the two groups was examined by means of a discriminant analysis, and the most important factors of separation were fixed by stepwise regression. The division into one of the two groups resulted from a quantity-frequency-index. An intake of an equivalent of 30 g of pure alcohol (100% alcohol) per drinking situation was sufficient to result in Blood Alcohol Concentration (BAC) of 1% in a woman of mean body weight. Subjects who consumed 30 g of pure alcohol (100% alcohol) or more per drinking situation were characterized as heavy consumers, those who took in a lower amount as light consumers. The two groups could be well separated by variables which had shown in other investigations, significant relationships to alcoholism or which had a strong association with heavy consumption in our own inquiry. By a direct discriminant analysis, a mean-value of 88.5% of the subjects could be classified in the correct group. The variability of 56% was explained by this procedure (P less than 0.001). In a stepwise regression, a mean-value of 80.5% of the subjects could be classified, and 41% of the variability was explained (P less than 0.001). A set of 20 variables was found to increase the discrimination, all of them significant with P less than 0.01. The variables entered into this discrimination analysis can partly be interpreted on the basis of the examination of the univariate evaluation as variables with a function of risk on one side or protection on the other.