There is good evidence that smoking is a marker for high steroid hormo
ne concentrations (at least at the time that smoking is initiated). Th
is would explain the finding that smoking is associated with dizygotic
, though not monozygotic, twinning. The notion that smoking and other
elective behaviours (e.g. drinking alcohol, opting for vasectomy and u
se of oral contraception) are markers or indices of high hormone conce
ntrations may have widespread repercussions for cross-sectional epidem
iological studies of such risk factors for diseases thought to be part
ially caused by high (e.g. prostatic and breast cancers) or low (e.g.
rheumatoid arthritis) concentrations of these hormones.