Although the relation between serum total cholesterol and coronary hea
rt disease is well established, the relation with mortality from non-c
oronary disease is controversial. Inverse relations of serum cholester
ol with hemorrhagic stroke and cancer have stimulated the examination
of cholesterol-non-coronary mortality associations. The population sur
veyed is 12,187 men and women aged 40-69 years living in Yao City, a s
uburb of Osaka, who undertook baseline examinations between 1975 and 1
984 and had no history of stroke and coronary heart disease at baselin
e. The subjects were followed on average 8.9 years until the end of 19
88 using systematic mortality surveillance. During the follow-up, ther
e were 343 deaths, comprising 170 cancer deaths (International Classif
ication of Death 9th edition: ICD-9, 140-239), 21 coronary heart disea
se deaths (ICD-9, 410-414), 67 other cardiovascular deaths (ICD-9, 390
-458 excluding 410-414), and 85 non-cardiovascular, non-cancer deaths.
There was a significant inverse association of serum cholesterol with
total and cancer mortality for men, and no significant association fo
r women. The cholesterol-disease association, although not significant
, was positive for coronary heart disease and other cardiovascular dis
ease deaths, and inverse for non-cardiovascular, non-cancer deaths in
both sexes. The inverse association of serum cholesterol with total an
d cancer mortality for men remained significant after controlling for
age, job classification, hypertension category, usual alcohol intake,
cigarette smoking, and relative weight index. A 34 mg/dl (1 SD) lower
serum cholesterol level was associated with a 21% (95% confidence inte
rval (CI) = 4-2%) higher risk of total mortality and a 25% (95% CI int
erval = 2-56%) higher risk of cancer death. These inverse associations
were stronger in current heavy drinkers (greater than or equal to 56
g ethanol per day) than in the other men. The inverse association betw
een serum cholesterol and cancer was also significant when deaths with
in the first 5 years were excluded from the analysis. Wether the inver
se association is explained by other factors or whether this associati
on is causal requires further investigation.