1. sure/B-Pb regression coefficients by 15 and indicator of actual alcohol use than alleged
2. 23% respectively. alcohol consumption (for which voluntary The strong association of B-Pb concen- under-reporting cannot be excluded).
3. tration to alcohol consumption prompted According to this B-Pb would be only an
4. us to analyze the B-Pb/blood pressure rela- indicator of alcohol intake and the effect
5. tion separately in drinkers and non- on blood pressure would be actually due to