Anonymous versus identifiable questionnaires in drug usage surveys.