A comparative study of two methods for the measurement of alcohol consumption in the general population

One of the major methodological problems in measuring alcohol consumption in a general population and in selected groups is underreporting.

The present study is based on a general health questionnaire survey of a random sample of about 4000 adults aged 20-74 years in an inner-city area in each of two major Swedish cities in 1991. The questionnaire included items both about alcohol consumption frequency and the usual amount of intake--the commonly used quantity-frequency (QF) method--and other questions about the consumption during work-days and weekends during a 'normal week'--the period-specific normal week (PSNW) method.

With a few exceptions, the reported mean consumption and the proportion of high consumers was higher with the latter approach, irrespective of sex, age, socio-demographic factors, smoking and health status, i.e. for variables which are commonly used as confounders or effect modifiers. The differences between the methods was greater among women. The internal non-response rate was higher with the PSNW method but the non-responders had a comparatively low consumption, when measured with the QF method.

The PSNW method has higher validity and greater precision for the measurement of alcohol consumption and, thus, is superior to the QF method. The sex differences are notable and warrant further studies focusing on sex-related modes of answering.