Correlations between biological markers and alcohol intake as measured by diary and questionnaire in men.

Alcohol intake was measured in 54 men aged 32-45 who were moderate or heavy drinkers--first by questionnaire and then more thoroughly by diary. Blood was drawn for the measurement of suggested biological markers of alcohol intake after the completion of the questionnaire and at the end of two consecutive diary periods. Partial correlations, controlling for smoking, exercise and age, between daily alcohol intake and markers were low and increased only slightly when diary data were analyzed instead of questionnaire data. No improvement was observed when mean intake per drinking day or the highest daily intake during a diary period was substituted in the analyses for mean daily intake. Significant associations (p less than .05) were found for mean cell volume (MCV), serum high-density lipoprotein cholesterol (HDL) and alkaline phosphatase (AFOS), but not for serum gamma-glutamyl transferase (GGT), the HDL: cholesterol ratio or ferritin. Correlations between intake reported by diary and intake predicted by a multivariate model with MCV, HDL, AFOS and GGT as regressors were significant (p less than .05) and higher than any of the bivariate correlation coefficients but were low (r = .30-.31). Low correlations seem to result from the inherent weakness of the suggested markers rather than from inaccuracies in the measurement of alcohol intake.