ABSTRACT
Evaluation of relationships between alcohol intake and health outcomes is hampered by dependence on study participants' self-reports of drinking and the lack of a simple biochemical marker which can validate long-term exposure. Researchers have been understandably cautious about accepting the validity of people's reports of their drinking habits because social pressure may lead some who drink excessively to under-report their drinking or even present themselves as nondrinkers. This type of error could lead to underestimation of the level of consumption at which risk is increased or to the appearance of a J-or U-shaped relationship in which the denial of high-risk heavy drinkers causes them to be erroneously placed with abstainers or very light drinkers (MacMahon, 1987; Marmot & Brunner, 1991). Studies cited in this paper generally considered this possibility in analyses and/or discussion of results. Most investigators used quantity-frequency type questionnaires completed by the participant in their home or a health care setting. Repeated assessments and/or correlation of alcohol consumption with biochemical markers (such as y-glutamyltransferase (GGT) or HDL-cholesterol) suggest these measures have acceptable reliability and validity (Fuchs et al., 1995; Rimm et al., 1991).
